Comparative and Diachronic Perspectives on Romance Syntax [1 ed.] 9781527509498, 9781527504011

The volume brings together fifteen papers focusing on the morphosyntax of different Romance varieties. It is based on pa

190 100 3MB

English Pages 463 Year 2018

Report DMCA / Copyright

DOWNLOAD FILE

Polecaj historie

Comparative and Diachronic Perspectives on Romance Syntax [1 ed.]
 9781527509498, 9781527504011

Citation preview

Comparative and Diachronic Perspectives on Romance Syntax

Comparative and Diachronic Perspectives on Romance Syntax Edited by

Gabriela Pană Dindelegan, Adina Dragomirescu, Irina Nicula and Alexandru Nicolae

Comparative and Diachronic Perspectives on Romance Syntax Edited by Gabriela Pană Dindelegan, Adina Dragomirescu, Irina Nicula and Alexandru Nicolae This book first published 2018 Cambridge Scholars Publishing Lady Stephenson Library, Newcastle upon Tyne, NE6 2PA, UK British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library Copyright © 2018 by Gabriela Pană Dindelegan, Adina Dragomirescu, Irina Nicula, Alexandru Nicolae and contributors All rights for this book reserved. No part of this book may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without the prior permission of the copyright owner. ISBN (10): 1-5275-0401-8 ISBN (13): 978-1-5275-0401-1

A Festschrift for Martin Maiden in honour of his 60th birthday

TABLE OF CONTENTS

Introduction ................................................................................................. 1 Part I: The Nominal Domain Chapter One ................................................................................................. 5 Inversion and Nominal Ellipsis: A Special Type of Nominalization in Romanian GABRIELA PANĂ DINDELEGAN Part II: Pronominal Clitics and Verb Movement Chapter Two .............................................................................................. 23 Parallels in Clausal and Nominal Structures: Romanian Clitic Placement ADAM LEDGEWAY Chapter Three ............................................................................................ 53 Syntactic Variation in Two Sister Languages: A Study of Word Order in Old French and Old Occitan SAM WOLFE Chapter Four .............................................................................................. 85 Syntactic Archaisms Preserved in a Contemporary Romance Variety: Interpolation and Scrambling in Old Romanian and Istro-Romanian ADINA DRAGOMIRESCU and ALEXANDRU NICOLAE Chapter Five ............................................................................................ 117 The Clitic Doubling Cycle: A Diachronic Reconstruction JORGE VEGA VILANOVA, SUSANN FISCHER and MARIO NAVARRO Chapter Six .............................................................................................. 135 A History of Personal Subject Pronouns in Milanese in Comparison with Other Northern Italian Dialects MASSIMO VAI

viii

Table of Contents

Chapter Seven.......................................................................................... 171 On the Ethical Dative in Old and Modern Romanian MIHAELA TĂNASE-DOGARU and CAMELIA U‫܇‬URELU Part III: Complementisers, Relative Clauses, and Coordination Chapter Eight ........................................................................................... 191 Towards a Microparameter C-Hierarchy in Italo-Romance VALENTINA COLASANTI Chapter Nine............................................................................................ 227 The Complementiser System of Aromanian (South Albania) M. RITA MANZINI and LEONARDO M. SAVOIA Chapter Ten ............................................................................................. 255 The Syntax of Ibero-Romance Quotation ALICE CORR Part IV: Finite and Non-Finite Verb Forms and Auxiliary Selection Chapter Eleven ........................................................................................ 289 Empirical and Analytical Problems of the Romance Inflected Infinitive: The State of the Art KIM GROOTHUIS Chapter Twelve ....................................................................................... 325 Old Romanian Finite Gerunds and Infinitives DANA NICULESCU Chapter Thirteen ...................................................................................... 361 Instability and Change: A Parametric Approach to Barese Auxiliary Selection LUIGI ANDRIANI Part V: Particles Chapter Fourteen ..................................................................................... 401 The Romanian Interrogative Particle oare in a Comparative and Historical Perspective ION GIURGEA

Comparative and Diachronic Perspectives on Romance Syntax

ix

Chapter Fifteen ........................................................................................ 433 On the Marking of Negative Presupposition in Regional Varieties of Italian FEDERICA COGNOLA and NORMA SCHIFANO

INTRODUCTION

This book brings together fifteen papers focusing on the morphosyntax of Romance varieties. The papers tackle different theoretical issues concerning current linguistic theory (relevant both for comparative and diachronic approaches) including parameters, features and their hierarchical organization, grammaticalization, word order changes, and the level of verb movement in different varieties. All the papers included here were presented at the workshop bearing the same title held at the University of Bucharest in November 2015. We dedicate this book to Professor Martin Maiden in honour of his 60th birthday. In addition to his seminal contributions to Romance linguistics, Professor Maiden also undertook the challenging task of coediting recent reference works such as The Cambridge History of the Romance Languages (2010, 2013) and The Oxford Guide to the Romance Languages (2016), two comprehensive works which have reshaped the paradigm of Romance linguistics. More importantly, however, Professor Maiden is a very dear friend (and indeed one of the informal fathers) of the winter workshops in Bucharest (which accompany the Annual Conference of the Department of Linguistics). We believe that offering this book to him on the occasion of his 60th birthday is a fitting way of acknowledging his many significant contributions to Romance and Romanian linguistics and of thanking him for all his generosity. The Editors

PART I: THE NOMINAL DOMAIN

CHAPTER ONE INVERSION AND NOMINAL ELLIPSIS: A SPECIAL TYPE OF NOMINALIZATION IN ROMANIAN GABRIELA PANĂ DINDELEGAN “Iorgu Iordan ‒ Al. Rosetti” Institute of Linguistics [email protected]

1. Objectives The first section describes three syntactic structures of old Romanian, linked by common mechanisms of inversion in the order of constituents and nominal ellipsis: the genitival phrase, the partitive prepositional phrase and a combined genitive-incorporating partitive prepositional phrase, all occupying argumental positions, atypical for the syntactic composition of these phrases. The second part deals with a special type of nominalization and compounding (the type deaproapele (“the fellowman”) [de + Adv + Det]), whose derivation has features in common with the first three constructions, but which has gone a step further in the process of recategorization. In contrast to the first three structures, in this construction inversion and ellipsis obligatorily associate with the phenomenon of recategorization of prepositional phrase headed by de as a noun. This is a complex phenomenon: lexically, the incorporation of the preposition takes place, therefore a process of compounding ([PP de [AdvP aproape]] > [DP deaproapele]); grammatically, a process of nominalization/substantivization takes place, with all its syntactic and inflectional effects.

2. Constructions ‘linked’ by inversion and nominal ellipsis 2.1. In old Romanian, a genitive can occur as a subject and as an object obligatorily preceded by the genitive/possessive marker a(l) (1a–c) and

6

Chapter One

optionally accompanied by the floating quantifier toți “all” (1d). The possessive phrase, which has the same distribution like the genitival phrase, has an identical behaviour (1f). (1)

a.

b.

c.

d.

e.

şi [GEN-P ai lui]S pre el and AL.M.PL his DOM him nu-l priimiră (CC1.1567: 1v) not=CL.M.3SG.ACC welcome.PS.3PL “and his people did not receive him” aşa [pre [GEN-P ai lui]]DO so DOM AL.M.PL his nu-i va not=CL.M.PL.ACC AUX.FUT.3SG părăsi în nevoie (CC1.1567: 83v) in need leave.INF “so he will not leave his people in need” [GEN-P Ai domnilor-voastre]S vor AL.M.PL lordships=your AUX.FUT.3PL avea leage la domnu nostru have.INF trial at lord our şi la noi (DÎ.1595: CII) and at us “your lordship’s people will have a trial with our lord and with us” Şi se boteză and CL.REFL.ACC.3SG baptize.PS.3SG însuşi şi [GEN-P ai lui [toţi]]S himself and AL.M.PL his all aciiaş (CPr.1566: 78) here “and he and all his people were baptized here” Că, iată, dragii miei, because behold dear.DEF.VOC my [POS-Pal mieu]S apropie AL.M.SG my draws.near cătră mine (Ev.1642: 310) towards me “because, behold, my dear ones, my man is coming closer to me”

Inversion and Nominal Ellipsis

f.

g.

7

Eu sântu păstoriulu cela I am shepherd.DEF CEL.M.SG bunulu, şi cunoscu good.M.SG.DEF and know.1SG [POS-Pale saleşu]DO şi-su cunoscutu AL.F.PL his and=am know.PPLE [PP de [POS-Pale mealeşu]AL PREP OBJECT] of AL.F.PL my “I am the good shepherd and I know his things and I am known for my things” (CC2.1581: 487) [PP ca [NP neşte [GEN-P ale unoru like some AL.F.PL some striini] Ø]] au pre strangers have.3PL DOM eale (CC2.1581: 486) them “but they treat them like those of strangers”

The occurrence of this construction ‒ which is the result of ellipsis (the ellipsis of the head noun; for types of ellipsis, see Nicolae 2012; 2013), is facilitated by the frequent inversion of the order of constituents in old Romanian DPs, i.e. the genitive preceding the head noun (2a,b). (2)

a.

se CL.REFL.3SG

b.

duse go.PS.3SG [DP [GEN-P aik

de that lui]i robik ti] AL.M.PL his slaves lor avuţia them wealth.DEF

chemă call.PS.3SG şi deade and give.PS.3SG lui (CC1.1567: 107r) his “He went to call his slaves and gave them his fortune” Iară fraţii cei mici and brothers.DEF CEL.M.P little.PL cheamă-se [DP [GEN-P aik lui]i call=CL.REFL.3PL AL.M.PL his ucenicik ti] (CC2.1581: 41) apprentices “and the little brothers are called apprentices”

8

Chapter One

An extremely interesting example is (1g), in which the indefinite article neşte precedes a genitive, as an effect of head ellipsis and substitution by the genitive. 2.2. A similar phenomenon occurs in partitive PPs, which, in OR, admit the ellipsis of the quantified nominal head, whose effect is the occurrence of the prepositional partitive restriction in argumental position, subject (3a,b) or direct object (4a-d). In 16th century texts, the most frequent partitive preposition is de (3, 4), but also the other partitive prepositions (den/din “of”; dentre/dintre “of”), that introduce the partitive restriction, may occur as arguments, subjects (5a–c) and objects (6a–d). Usually, partitive prepositions select an NP whose head is in the plural (see (3a,b), (4a,b), (5a–c), (6c)); the selection of a singular form is also possible, with collective or mass nouns ((4c); (6a,b,d)); see also Pană Dindelegan (2016: 327). (3)

a.

b.

(4)

a.

b.

ştie tatălu vostru den knows father.DEF.NOM your from ceriu că trebuiaşte-vă [PP de heaven that need=CL.DAT.2PL of acealea]S (CC2.1581: 216) those “your heavenly father knows that you need those” după potopu au fostu [PP de after flood AUX.PERF.3PL been of toate bucatele]S (CC2.1581: 51) all foods “after the flood there were all kinds of food” Cum au fostu zăsă how AUX.PERF.3SG be.PPLE say.PPLE.F noao, [PP de toate legumile]DO us.DAT of all vegetables să mâncămu (CC2.1581: 51) SĂSUBJ eat.SUBJ.1PL “how we were told, to eat of all kinds of vegetables” [PP di toate]DO să cumpere of all.F.PL SĂSUBJ buy.SUBJ.3SG≡PL “to buy a bit of everything” (DÎ.1600: XXII)

Inversion and Nominal Ellipsis

c.

(5)

a.

b.

c.

(6)

a.

b.

9

vom mânca [PP de acest AUX.FUT.1PL eat.INF of this strâns]DO (Sind.1703: 100r) gathered “we shall eat of what we have gathered” au picat şi [PP den AUX.PERF.3PL fall.PPLE and of săimeni şi den căzaci]S Turkish.soldiers and of Cossacks “some of the Turkish soldiers and some of the Cossacks fell” (CLM.1700–50: 267v) s-au strânsu [PP din CL.REFL=AUX.PERF.3PL gather.PPLE of târgoveţi şi den slugile townsmen and of servants neguţitorilor]S (CLM.1700–50: 176v) traders.DEF.GEN “some of the townsmen and of the traders’ servants gathered” Iar unii dzicè că [PP dintre of and some said.3PL that dânşii]S l-au them CL.ACC.M.3SG=AUX.PERF.3SG≡3PL otrăvit (NL.~1750–66: 279) poison.PPLE “and some said that one of them poisoned him” să opreacă de ici SĂSUBJ stop.SUBJ.3SG≡3PL of here nainte [PP den bir]DO (DÎ.1600: XXX) on of tax “but they should keep a part of the tax from now on” să însoare pre Pătraşco-vodă şi SĂSUBJ marry DOM Pătraşco=prince and să-i dea [PP din SĂSUBJ=CL.DAT.3SG give.SUBJ.3SG of ruda împăratului]DO (DÎ.1600: XXXII) relative emperor.DEF.GEN “to marry prince Pătraşco and to give him one of the emperor’s relatives as a wife”

10

Chapter One

c.

d.

vindeţ [PP den averile sell.IMP.2PL of fortunes.DEF voastre]DO (Ev.1642: 285) your “sell a part of your fortunes” să-şi răscumpere SĂSUBJ=CL.REFL.DAT.3PL pay.back.3PL [PP den singele ce le that CL.DAT.3PL of blood.DEF făcuse săimeanii do.PLUPERF.3PL Turkish.soldiers.NOM lor]DO (CLM.1700–50: 187v) their “to pay back some of the blood of the Turkish soldiers that he had shed”

As in the case of the genitive, the placement of the prepositional partitive restriction in an argumental position is the result of ellipsis: the nominal head is elided (the head of the partitive construction). As in the case of the genitive, ellipsis is favoured by the frequent anteposition of the partitive restriction (7a–c): (7)

a.

b.

c.

[DP[PP dentr-a patra parte]i of=A fourth part giumătate ti] (DIR.A.I.1601: 21) half “a half of the forth part” mergând [DP [PP dintr-aceia]i going of=those unii ti] (CIst.1700–50: 17r) some.M.PL “some of those going” au şi peritu [DP [PP den AUX.PERF.3PL also died of munteani]i câţva ti] (CLM.1700–50: 229v) Wallachians some.M.PL “some of the Wallachians died”

2.3. A special type of ‘de’ partitive structure (see also Frâncu 1983; Giurgea 2013: 103; Pană Dindelegan 2016: 331) is the one that includes a genitive (8a) or a possessive phrase (8b,c). Examples like (9a–c) suggest

Inversion and Nominal Ellipsis

11

that the mechanism by which the partitive structure containing a genitive / a possessive occurs in argument position is the variation in word order resulting from the topicalization of the partitive structure. We can notice, in (8a,c), the coordination of an NP selected by a verb with a partitive prepositional phrase resulted of an ellipsis. (8)

a.

b.

c.

(9)

a.

când va cădea cineva when AUX.FUT.3SG fall.INF somebody.NOM în boală, [au [PP de [GENP ai in disease or of AL.M.PL casei]], au [DPmăcar şi house.DEF.GEN or even also streinii]]S (AAM.1713: 23v) strangers.DEF.NOM “when someone gets ill, either somebody in the family or even strangers” au luat [PP de-[POSP al AUX.PERF.3SG take.PPLE of=AL.M.SG nostru]]DO şi our and ne- au dat CL.DAT.1PL=AUX.PERF.3SG give.PPLE [PP de-[POSP al său]]DO of=AL.M.SG his “they took from what is ours and they gave some of his own” (DPar.1683: III/95v) strângându [[DP oşti] şi [PP de gathering armies and of [POSP ai săi]]]DO (CLM.1700–50: 171r) AL.M.PL his “gathering armies and some of his people” Şi periră acolo [DP [PP de-i and die.PS.3PL there of=AL.M.PL lu Por]I 4.000 de mii ti] LUI.GEN Por 4000 of thousand şi [DP [PP de-i lu and of=AL.M.PL LUI.GEN 6.000 tj] (A.1620: 33r) Alexandru]j Alexandru 6000 “And 4000 of Por’s people and 6000 of Alexander’s died there”

12

Chapter One

b.

c.

că lesne iau unii because easy take.3PL some.M.PL [PP de [GENP ale altora]]i [DP of AL.F.PL others.GEN cuvinte ti] words (CIst.1700–50: 39v) “because some take some of other people’s words easy” nu numai streini, ce şi not only strangers but also [DP [PP de [POSP a noştri]]i of AL.INV our moldoveni ti] Moldavians “not only strangers, but also some of our Moldavians” (NL.1750–66: 65)

The ellipsis of the nominal head is common to the three patterns (§§2.1; 2.2; 2.3), facilitated by the postposition of the head (see also Nicolae 2016: 568). There is a significant difference between the first (§2.1) and the last two patterns (§§2.2; 2.3): the first one occurs when there is an anaphoric component that conserves morphosyntactic features of the elided component (e.g. the possessive-genitival marker “al”), a structure that resembles the ‘cel’ nominal ellipsis structure (cel frumos ‘the beautiful one’; see Dragomirescu, Nicolae 2016), while the pattern with a partitive PP is characterized by the complete loss of the information of the elided component.

3. Adverb nominalizations; deaproapele (“the fellowman”), a special type of nominalization Many studies dedicated to old Romanian signal the frequency of nominalized adverbs (see, for example, Densusianu 1961 [1938]: 198, where the locatives (mai)susul “higher.DEF”, (mai)giosul “lower.DEF”, the tense adverbs apoia “afterwards.DEF”, apoii “then.DEF”, apoile “then.DEF”; astăzile “today.DEF”, the quantitatives multa “much.DEF”, destulul “enough.DEF” are listed as nominalized adverbs (10). In some studies on old Romanian, the discussion of adverb nominalization takes place in a larger context of adjective and participle nominalization (Stan 2012; 2013: 27–8; Dragomirescu, Nicolae 2016; the latter make a terminological distinction between nominalization, realized by attaching the enclitic determiner, and nominal ellipsis, realized with the determiner cel, a free

Inversion and Nominal Ellipsis

13

morpheme (noua vs cea nouă “the new one”; derepţii vs cei d(e)repţi “the righteous”). (10)

a.

b.

începu de la apoii start.PS.3SG from afterwards.PL.DEF până la întăii (CT.1560–1: 42v) until first.PL.DEF “he started from the last ones to the first ones” voru fi întâii AUX.FUT.3PL be.INF first.PL.DEF.NOM apoii şi apoii afterwards.DEF and afterwards.DEF.NOM întâii (CC2.1581 : 201) first.PL.DEF “the first ones will be the last and the last ones will be the first”

3.1. A syntactic pattern similar to the ones discussed under §2 is the one in which the nominal head is elided, when it is accompanied by an adverbial modifier of the noun, introduced by the preposition de and placed, just like in the other elliptical phrases, in argumental position (11a,b): (11)

a.

b.

Dumnedzău învaţă să agiutorim God teaches SĂSUBJ help.SUBJ.1PL [[fratele nostru]DO şi our and brother.DEF [deaproapele nostru cel fellowman.DEF our CEL.M.SG sărac]]DO (CazV.1643: 228r) poor “God teaches us to help our brother and our poor fellowman” Iară [de-aproapele nostru]S iaste tot and fellowman.DEF our is all omul ce-i trebuiaşte milă man that=CL.DAT.3SG need.3SG pity şi agiutoriu de la noi and help from us “and the fellowman is anyone that needs our pity and help” (CazV.1643: 344r)

14

Chapter One

One can notice that, in this context, a different de is used, which is not partitive de, but a functional de, specialised for introducing nominal modifiers. The construction is based on an adverbial modifier of the noun, of the type (12): (12)

a.

b.

era rudă de aproape lu was relative of close LUI.DAT Hristos (CC1.1567: 198r) Christ “he was a close relative of Christ” ca neşte fraţi de-aproape like some brothers of=close “like close brothers” (CazV.1643: 293r)

Usually, adverbial modifiers of the noun occur postnominally (12), but, like all the other constituents that are subordinate to the archaic NP, they can also occur prenominally (13a-c), a word order that facilitates the licensing of ellipsis. (13)

a.

b.

c.

carea o am which.DEF.ACC CL.ACC.F.3SG AUX.PERF.1SG arătat cătră acest show.PPLE towards this adevărat şî de aproape true and of close priiaten (SVI.~1670: 116r) friend “which I have shown to this true and close friend” întru ceale de-apoi in those of=afterwards dzâle (DPar.1683: I/4v) days “in those later days” Socotind prorocul acea judging prophet.NOM that de-apoi a ta, Hristoase, of=afterwards AL.F.SG your Christ.VOC venire (DPar.1683: III/109v) coming “as the prophet was thinking of your last coming, Christ”

Inversion and Nominal Ellipsis

15

3.2. Differently from the situations discussed above, in these contexts, inversion and ellipsis go one step further, being obligatorily accompanied by the phenomenon of recategorization of the de-PP phrase as a noun. This phenomenon is complex: lexically, it presupposes a process of Pincorporation, i.e. a process of compounding ([PP de [AdvP aproape]]) > [DP deaproapele]), resembling the one in dâns(ul) “he” (de + însu); grammatically, nominalization takes place, with all of its syntactic and inflectional effects. The syntactic diagnostics which signal the loss of autonomy of the preposition and its incorporation as part of a lexical word in this structure are the following: – occurrence of the structure as a S (see (11b), (14a,b)) and a DO (see (11a)); see also its occurrence as an IO (14c), where it can be coordinated with a prototypical indirect object); as a DO, deaproapele is constructed, like all personal nouns of this period, either with (15b,c) or without the differential marker pre (15a); – occurrence in the genitive (16a,b); – association with another preposition ‒ any lexical preposition (cu “with”, cătră “towards”, spre “towards”) (17a,b); a decisive argument is its association with another instance of de, the first one being selected by the head (17c): (14)

a.

b.

c.

Cine iaste [de-aproapele who is of=close.one.DEF.NOM nostru]S? (CazV.1643: 292v) our “Who is our fellowman?” A treia arată că AL.F.SG third shows that [de-aproapele]S nu să cheamă of=close.one.DEF not CL.REFL call.3SG omul (…) (CazV.1643: 341v) man.DEF “The third one shows that our fellowman is not called a man” de folos iaste [[lui Dumnedzău]IO of use is LUI.DAT God [de-aproapelui şi [şie]OI şi and himself and of=close.one.DEF.DAT

16

(15)

Chapter One

a.

b.

c.

(16)

a.

b.

(17)

a.

b.

c.

său]IO]IO (CazV.1643: 208r) his “it is useful to God and to himself and to his fellowman” Să iubeşti [de-aproapele tău]DO SĂSUBJ love.SUBJ.2SG of=close.one.DEF your ca sângur pre tine as yourself DOM you “Thou shalt love thy neighbour as thyself” (ŞT.1644: 58) Pentru ce tu baţ for what you beat.2SG [pre de-aproapele]DO? (DPar.1683: III/35v) DOM of=close.one.DEF “Why are you beating your fellowman?” Sau [pre deaproapele său]DO or DOM of.close.one.DEF his îl pizmuiaşte (CPV.~1705: 62v) CL.ACC.M.3SG envy.3SG “or he envies his fellowman” să apuce hotarul SĂSUBJ grasp.SUBJ.3SG border.DEF.ACC [GENPdeaproapelui său] (Prav.1646: 54) of.close.one.DEF.GEN his “in order to take over the land of his fellowman” glasului [GEN-P de-aproapelui] of=close.one.DEF.GEN voice.DEF.GEN≡DAT “(to) the fellowman’s voice” (DPar.1683: III/2r) liubovul cătră Dumnezeu şi love.DEF towards God and [PP cătră [DP deaproapele său]] his towards of.close.one.DEF “the love for God and for his fellowman” (VRC.1645: 2v) va vătăma foarte mult AUX.FUT.3SG hurt.INF very much [PP spre [DP deaproapele fieşcui]] towards of.close.one.DEF everyone.GEN “he will hurt everyone’s fellowman a lot” (Prav.1646: 170) nu i să not CL.DAT.3SG CL.REFL.3SG

Inversion and Nominal Ellipsis

17

face milă [PP de becomes pity of [DP de-aproapele său]] (DPar.1683: III/20r) of=close.one.DEF his “he will not have pity for his fellowman” As an inflectional cue for nominalization, we mention the association with the enclitic article (18a,b): (18)

a.

b.

iubeaşte [de-aproapele tău]DO love.IMP.2SG of=close.one.DEF your “love your fellowman” (CazV.1643: 274r) Tot însul [de-aproapelui]IO all man.NOM of=close.one.DEF.DAT s-agiute (DPar.1683: III/8r) SĂSUBJ=help.SUBJ.3SG “all men should help their fellowmen”

As a noun, deaproapele does not have a plural form; other nominalizations de(n)apoiuri (of-after-PL “the things that follows”), alsăuri (al-his(POSS.ADJ)-PL “his belongings”) have a plural form, marked by the inflexion –uri, usually used as a nominalization device.1 3.3. The nominalization deaproapele belongs to the 17th century, with a peak in the period 1640–1700. After 1700, its occurrence is ever increasingly rarer, as its nominal form is limited to the non-prepositional construction of the adverb (aproapele). The form deaproapele is predominant in Moldova, in texts such as CazV.1643, ŞT.1644, Prav.1646, DPV.1673, DPar.1683; for example, in just one text (CazV.1643), the de-pattern occurs 14 times as a DO, 4 times 1

Adverbial nominalization, irrespective of the nominalization pattern, selects common inflectional markers: the enclisis of the definite determiner and the neuter plural marker -uri: lăuntruri (inside.PL “insides”, CP1.1577: 14r); nontrurile (inside.PL.DEF ‘the insides’, DPar.1683: IV.48v); napoiurile (back(wards).PL.DEF “the backward parts”, DPar.1683: IV.53r)). It is interesting that other special nominalizations, some of which are accidental, have an -uri plural form, showing that -uri, even in his earliest occurrences, also had the supplementary function of nominal categorizer; see, for example, the nominalization of the possessive: alsăurile (al-his(POSS.ADJ)-PL.DEF) “(the) alsăuri (al-his(POSS.ADJ)-PL), belongings” (DPar.1683: III.127r; AD.1722–5: 121r); for the role of the inflexion -uri as a nominal categorizer, see Pană Dindelegan (2002: 38–9; 2009: 23).

18

Chapter One

as a S, 3 times as an IO, 5 times as a PP, one time as a predicative, 4 times as a genitive. Without being as frequent, the form deaproapele is not completely absent from Wallachia (a few example occur in Mărg.1691 or in NÎnv.~1700) or from Transylvania (see, in CDicţ.1691–7: 413, synonyms rudă “relative”, priiatin “friend”, deaproapele “the close one”; see also CPV.~1705: 62v).

4. Conclusions Four old Romanian nominal constructions show the same mechanism of inverted word order and ellipsis, which makes possible the occurrence of non-prototypical constituents as arguments (the genitive phrase, the partitive prepositional object, the partitive prepositional object that incorporates a genitive phrase and an adverbial prepositional modifier; see also the placement of the genitive phrase on the right-hand side of the indefinite determiner neşte (1g)). Only one of these constructions (a prepositional modifier of the noun) undergoes a further change, taking part in a process of compounding (the incorporation of the preposition de) and of nominalization (the result is deaproapele “the close one.DEF”). In the 17th century, the pattern deaproapele [de+Av+Art]N has a high frequency in Moldova; rarely, it occurs in Wallachia and in Transylvania; afterwards, the structure deaproapele becomes obsolete, being replaced by its nominalized correspondent aproapele, without preposition ([Av+Art]N), still currently attested.

Corpus A.1620 = Alexandria. Ed. F. Zgraon, Bucharest, Fundaţia Naţională pentru Ştiinţă şi Artă, 2005 (Cele mai vechi cărţi populare în literatura română, 11). AAM.1713 = Antim Ivireanul, Aşezământul Mănăstirii Antim. Ed.: Antim Ivireanul, Opere, ed. G. Ştrempel, Bucharest, Minerva, 1972, 324–46. CazV.1643 = Varlaam, Cazania, ed. J. Byck, Bucharest, Fundaţia pentru Literatură şi Artă, 1943. CC1.1567 = Coresi, Tâlcul Evangheliilor. Ed.: Coresi, Tâlcul evangheliilor şi molitvenic românesc, ed. V. Drimba, Bucharest, Editura Academiei, 1998, 31–187. CC2.1581 = Coresi, Evanghelie cu învăţătură (Braşov). Ed. S. Puşcariu, Al.Procopovici: Diaconul Coresi, Carte cu învăţătură (1581), vol. I, Textul, Bucharest, Socec, 1914. CDicţ.1691–7 = Theodor Corbea, Dictiones Latinæ cum Valachica interpretatione. Ed. A.-M. Gherman, vol. I, Cluj-Napoca, Clusium, 2001.

Inversion and Nominal Ellipsis

19

CIst.1700–50 = Constantin Cantacuzino, Istoria Ţării Româneşti. Ed.: Istoria Ţărâi Rumâneşti atribuită stolnicului Constantin Cantacuzino, ed. O. Dragomir, Bucharest, Editura Academiei Române, 2006, 145–202. CLM.1700–50 = Miron Costin, Letopiseţul Ţărâi Moldovei. Ed.: M. Costin, Opere, ed. P. P. Panaitescu, Bucharest, Editura de Stat pentru Literatură și Artă, 1958, 41–201. CPr.1566 = Coresi, Apostol. Ed. I. Bianu, Texte de limbă din secolul XVI, IV, Lucrul apostolesc tipărit de diaconul Coresi la 1563, Bucharest, 1930. CPV.~1705 = Theodor Corbea, Psaltirea în versuri. Ed. A.-M. Gherman, Bucharest, Editura Academiei Române, 2010. DÎ = Documente şi însemnări româneşti din secolul al XVI-lea, text stabilit şi indice de Gh. Chivu, M. Georgescu, M. Ioniţă, Al. Mareş, Al. RomanMoraru, Bucharest, Editura Academiei Române, 1979. DIR.A.I = Documente privind istoria României, veacul XVII, A, Moldova, Bucharest, Editura Academiei I. DPar.1683 = Dosoftei, Parimiile preste an, Iaşi, 1683, ed. M. Ungureanu, Iaşi, Editura Universității „Al. I. Cuza”, 2012. DPV.1673 = Dosoftei, Psaltirea în versuri (Uniev). Ed.: Dosoftei, Opere, 1, Versuri, ed. N. A. Ursu, Iaşi, Mitropolia Moldovei și a Sucevei, 1974. Ev.1642 = Evanghelie învăţătoare (Govora, 1642). Ed. A.-M. Gherman, Bucharest, Editura Academiei Române, 2011. Mărg.1691 = Mărgăritare . Ioan Gură de Aur, Mărgăritare, ed. R. Popescu, Bucureşti, Libra, 2001, p. 11–493. NÎnv.~1700 = Învățăturile lui Neagoe Basarab către fiul său Teodosie. Ed. F. Moisil, D. Zamfirescu, Bucharest, Minerva, 1971, p. 125–352. NL.~1750–66 = Ion Neculce, Letopiseţul. Ed.: Ion Neculce, Letopiseţul Ţării Moldovei şi O samă de cuvinte, ed. I. Iordan, Bucharest, Editura de Stat pentru Literatură și Artă, ed. a II-a, 1959, p. 31–388. PO.1582 = Palia de la Orăştie. Ed. V. Pamfil, Bucharest, Editura Academiei R.S.R, 1968. Prav.1646 = Carte românească de învăţătură (Iaşi). Ed.: Carte românească de învăţătură. 1646, ed. Colectivul pentru vechiul drept românesc condus de acad. A. Rădulescu, Bucharest, Editura Academiei R.P.R., 1961, p. 33–106. Sind.1703 = Sindipa. Ed. M. Georgescu, Bucharest, Minerva, 1996 (Cele mai vechi cărţi populare în literatura română, 1), p. 249–315. ŞT.1644 = Şeapte taine a besearecii, Iaşi, 1644. Ed. I. Mazilu, Iaşi, Editura Universitatii „Alexandru Ioan Cuza”, 2012. SVI.~1670 = Varlaam şi Ioasaf. Ed.: M. Stanciu Istrate, Reflexe ale medievalităţii europene în cultura română veche: Varlaam şi Ioasaf în cea mai veche versiune a traducerii lui Udrişte Năsturel, Bucureşti, Editura Muzeului Național al Literaturii Române, 2013, p. 82–325. VRC.1645 = Varlaam, Răspunsul împotriva catihismusului calvinesc. Ed.: Varlaam, Opere, Răspunsul împotriva catihismului calvinesc, ed. M. Teodorescu, Bucharest, Minerva, 1984, p. 143–230.

20

Chapter One

References Densusianu, O., 1961 [1938], Istoria limbii române, Vol. II, Secolul al XVI-lea, (ed.) J. Byck, Bucharest, Editura Ştiinţifică, 1961. Dragomirescu, A., A. Nicolae, 2016, “L’ellipse nominale avec article défini de l’ancien roumain au roumain moderne: Le cas du participe passé”, in: E. Buchi, J.-P. Chauveau, J.-M. Pierrel (eds), Actes du XXVIIe Congrès international de linguistique et de philologie romanes (Nancy, 15-20 juillet 2013), 3 volumes, Strasbourg, Société de linguistique romane/ÉliPhi, 155– 170. Frâncu, C., 1983, “Vechimea şi răspândirea construcţiei partitive de tipul un prieten de-al meu”, Limba română, XXXII, 1, 15–23. Giurgea, I., 2013, Originea articolului posesiv-genitival al şi evoluţia sistemului demonstrativelor în română, Bucharest, Editura Muzeului Naţional al Literaturii Române. Ledgeway, A., 2009, Grammatica diacronica del dialetto napoletano, Tübingen, Max Niemeyer Verlag. Nicolae, A., 2012, “Two Sources for ellipsis in Universal Grammar”, International Workshop ‘ELLIPSIS 2012: crosslinguistic, formal, semantic, discursive and processing perspectives’. Nicolae, A., 2013, Types of Ellipsis in Romanian, PhD Dissertation, University of Bucharest & University of Cambridge. Nicolae, A., 2016, “Word order and configurationality”, in: G. Pană Dindelegan (ed.), The Syntax of Old Romanian, Oxford, Oxford University Press, 563-575. Pană Dindelegan, G., 2002, “Formaţii substantivale recente şi rolul „clasificatorilor” în actualizarea lor contextuală”, in: G. Pană Dindelegan (ed.), Aspecte ale dinamicii limbii române actuale, Bucharest, Editura Universităţii din București, 31–46. Pană Dindelegan, G., 2009, “Trăsături ale flexiunii substantivului în româna actuală”, in: G. Pană Dindelegan (ed.), Dinamica limbii române actuale— aspecte gramaticale şi discursive, Bucharest, Editura Academiei Române, 3-33. Pană Dindelegan, G., 2016, “The partitive phrase”, in: G. Pană Dindelegan (ed.), The Syntax of Old Romanian, Oxford, Oxford University Press, 323–331. Stan, C., 2012, “Aspecte diacronice ale sintaxei articolului definit în limba română”, in: R. Zafiu, A. Dragomirescu, A. Nicolae (eds), Limba română: direcţii actuale în cercetarea lingvistică, Bucharest, Editura Universităţii din Bucureşti, 239–244. Stan, C., 2013, O sintaxă diacronică a limbii române vechi, Bucharest, Editura Universităţi din Bucureşti.

PART II: PRONOMINAL CLITICS AND VERB MOVEMENT

CHAPTER TWO PARALLELS IN CLAUSAL AND NOMINAL STRUCTURES: ROMANIAN CLITIC PLACEMENT ADAM LEDGEWAY Downing College, University of Cambridge [email protected]

1. Introduction A well known idiosyncrasy of modern Romanian syntax (cf. Sandfeld and Olsen 1936: 97; Lombard 1974: 128f.; Dobrovie-Sorin 1994: 22; Legendre 2000: 232; Motapanyane and Alboiu 2000: 9f.; Dragomirescu 2013: 193f.; Vasilescu 2013: 387) is the behaviour of the feminine singular accusative clitic o (< ILLAM). In contrast to the generalized proclisis found with all other pronominal clitics such as the masculine singular accusative clitic (î)l (< ILLUM; cf. 2a-g) and with o itself in conjunction with synthetic verb forms (cf. 1a), the feminine singular accusative clitic exceptionally occurs enclitic to the participle/infinitive in conjunction with the indicative analytic perfect and substandard analytic pluperfect (1b), the conditional (perfect) (1c), and the colloquial future/presumptive (perfect) in oi, ei,... (cf. 1d), as well as optionally—or substandardly in some cases—with the (perfect) infinitive (1e), the analytic literary future in voi, vei,… (1f), and the perfect subjunctive (cf. 1g). (1)

a.

b.

O opresc(*-o) her= I.stop(=her) “I stop her” (*O) am oprit-o // (*O) her=I.have stopped=her her= fost oprit-o been stopped=her “I have stopped her / I had stopped her”

am I.have

24

Chapter Two

c.

d.

e.

f.

g.

(2)

a.

b.

c.

(*O) aș opri-o // (*O) her= I.would stop.INF=her her= aș fi oprit-o I.would be stopped =her “I would stop her / I would have stopped her” (*O) oi opri-o // (*O) oi her=I.will stop.INF=her her= I.will fi oprit-o be stopped =her “I will stop her / I will have stopped her” Faptul de a (o) opri(-o) // fact.the of to her= stop.INF=her Faptul de a (o) fi oprit(-o) fact.the of to her= be stopped=her “the fact of stopping her / the fact of having stopped her” (O) voi opri(??-o) her= I.will stop.INF=her “I will stop her” Nu cred să (o) fi not I.believe thatSBJV her=be oprit(??-o) stopped=her “I don’t believe that I/you/(s)he/we/they have/has stopped her” Îl opresc him= I.stop “I stop him” Lam oprit(*u-l)// him= I.have stopped=him Lam fost oprit(*u-l) him= I.have been stopped=him “I have stopped him / I had stopped him” Laș opri(*-l) // him= I.would stop.INF=him Laș fi oprit(*u-l) him= I.would be stopped =him “I would stop him / I would have stopped him”

Parallels in Clausal and Nominal Structures: Romanian Clitic Placement

d.

e.

f.

g.

25

Loi opri(*-l) // him= I.will stop.INF=him Loi fi oprit(*u-l) him= I.will be stopped=him “I will stop him / I will have stopped him” Faptul de a -l opri(*-l) // fact.the of to =him stop.INF=him Faptul de a -l fi fact.the of to =him be oprit(*u-l) stopped=him “the fact of stopping him / the fact of having stopped him” Îl voi opri(*-l) him= I.will stop.INF=him “I will stop him” Nu cred să -l fi not I.believe thatSBJV =him be oprit(*u-l) stopped=him “I don’t believe that I/you/(s)he/we/they have/has stopped him”

Available descriptions of the phenomenon, however, provide a less than consistent picture of usage in this area which ultimately points to the conclusion that we are in part witnessing a change in progress or, at the very least, a relatively recent change in the grammar, the effects of which are still not yet distributed uniformly across all speakers. Indeed, discrepancies between different descriptions seem to highlight a combination of partly converging syntactic and phonological forces at play which have produced some ongoing analogical patterns of widening and narrowing in the distribution of the feminine singular accusative pronoun. By way of example, consider the summary generalizations extracted from a number of representative sources and grammars illustrated in Table 1.

Chapter Two

Sandfeld and Olsen (1936: 97) Lombard (1974: 128) Dobrovie-Sorin (1994: 22) Legendre (2000: 232) Motapanyane and Alboiu (2000: 9f.) Monachesi (2005: 169) Dragomirescu (2013: 193f.) Rîpeanu Reinheimer et al. (2013: 251) Vasilescu (2013: 387) Sarlin (2014: 144) Nicolae (2015:§3.2) Nicolae and Niculescu (2016: 61) E E E E E E E

E E E E E E E

PrsPrf E E E E E

E

E

E E

Vowel-Initial Verb Cnd CndPrf PopFut E E E E E E E E E

Table 1: Placement of modern Romanian o (E(nclitic); P(roclitic); E/P = optional; ? = not preferred)

26

E

PopFutPrf E

E/P

Inf

P P P

E/P E/P ?E/P

P

E/P

Consonant-(/Vowel-)Initial Verb PrfInf LitFut LitFutPrf PrfSbjv E/?P E E P E P E/?P

Parallels in Clausal and Nominal Structures: Romanian Clitic Placement

27

From Table 1 it is immediately obvious that the distribution of enclisis with the clitic o is much more firmly established in those periphrases and contexts in which the first verb, the potential verbal host, begins with a vowel. Consequently, in the first five columns of Table 1, ranging from the present perfect to the popular future perfect, we see that enclisis to the participle or infinitive is consistently reported by all authors as the only option. By contrast, in contexts in which the auxiliary or lexical verb does not (necessarily) begin with a vowel, namely in the shaded part of the table containing the final five columns, we see that the picture is less clear, in that enclisis is variously claimed to be the only option (e.g. perfective infinitive), optional (e.g. infinitive), or preferred but not entirely to the exclusion of proclisis (literary future). In other cases, the various descriptions contradict one another. For instance, Dobrovie-Sorin (1994: 22) interprets proclisis as marginal with the literary future, while Rîpeanu Reinheimer, Tasmowski and Vasilescu (2013: 251) judge enclisis to be marginal in the same context. Similarly, Sandfeld and Olsen (1936: 97) and Lombard (1974: 128) both consider enclisis obligatory with the literary future perfect, whereas Nicolae and Niculescu (2016: 61) only recognise proclisis as grammatical in this context. Despite these differences, what does emerge clearly from Table 1 is that enclisis of o is most robustly established in contexts where the potential verbal host begins with a vowel, whereas in all other contexts we variously find a combination of enclisis and proclisis. It is therefore unsurprising that many analyses, as we shall see, have ascribed the enclisis of o to phonological reasons. In what follows we therefore aim to provide answers for at least four related questions. First, what is special about the feminine singular accusative clitic o that exceptionally gives rise to enclitic sequences such as Aux+V=cl which are not available to other object clitics such as the masculine singular accusative clitic (î)l? Second, how many clitic positions or sites are there within the clause, and how are the observed enclitic (cf. 1b) and proclitic (cf. 2b) positions related? Third, why is there a growing (diachronic) tendency, at least in certain registers, for enclisis of o to be preferred over proclisis in contexts such as (1e-g)? Finally, if the observed proclisis-enclisis alternation is, at least in part, driven by phonological considerations, then why does the clitic o invariably surface proclitic to finite lexical verbs (cf. 1a), irrespective of whether the verbal host is vowel- or consonant-initial?

28

Chapter Two

2. Phonological accounts We begin our investigation by considering those accounts which endeavour to explain the obligatory enclisis of o in contexts such as (1b-d) by appealing to phonological ill-formedness. Indeed, such a view is not uncommon in the literature, witness the representative selection of claims presented in (3a-d): (3)

a. b.

c.

d.

“il est évident qu’on ne pouvait admettre *o oi, *o om, *o or” (Lombard 1974: 129) “When the auxiliary verb begins with a vowel, the feminine accusative third person singular form o is placed after the main verb” (Sarlin 2014: 144) “the fact that o ‘her’ is postverbal […] cannot be related to a syntactic peculiarity but most probably involves phonological processes. It is indeed difficult to understand why o is necessarily postverbal on the basis of its [+fem] feature; its phonological “weakness” is instead likely to be relevant.” (Dobrovie-Sorin 1994: 73) “the reason for the ill-formedness […] is of a phonological nature. The clitic o cannot precede auxiliary which begins with a vowel. In the case of the future paradigm, the auxiliary begins with a consonant and the clitic o can occur either after the infinitive […] or in front of the auxiliary” (Monachesi 2005: 169)

What all such approaches have in common is their claim that the illformedness of proclisis in (1b-d) would result in a PF violation (cf. also Bošković 2001). In particular, it is claimed that the combination of the feminine singular accusative clitic and a vowel-initial auxiliary is ruled out on account of a language-particular phonological constraint which excludes sequences of two immediately contiguous non-front vowels, as apparently revealed by the non-shaded section of Table 1 above. However, these purely phonological accounts prove entirely misplaced on a number of fronts.

2.1 Synchronic evidence From a synchronic perspective, for instance, we see that the supposedly illicit phonological sequence involving two non-front vowels assumed in (3a-d) is simply not supported by the empirical evidence,

Parallels in Clausal and Nominal Structures: Romanian Clitic Placement

29

inasmuch as the mid-back vowel is regularly found immediately adjacent to all vowel combinations (e.g. [o, o̯] + [o-, a-, i-, oj-, ej-, aj, om-, am-, ar-, or-…]) both within words and across words, witness the representative examples in (4a-b). (4)

a.

b.

o albină, o aducem, a= bee her= we.bring o ardeau, o her= they.were.burning a= poartă… gate o opresc, o omor, o her= I.stop her= I.kill a= orație, o oală… speech a= pot

Moreover, we also observe that the ill-formedness of examples such as (1b) in which proclisis of the feminine singular accusative clitic to the first person singular auxiliary am “I have” is said to be excluded on phonological grounds is immediately disproven by examples such as (5), in which exactly the same phonological sequence with copular am “I have” proves entirely grammatical (cf. Rîpeanu Reinheimer, Tasmowski and Vasilescu 2013: 215f.): (5)

O am(*-o) her= I.have=her “I have her/it with me”

cu with

mine me

Significantly, the phonological approach also fails to take account of (optional) enclisis (cf. 1e-g) in conjunction with the (perfective) infinitive (6a-b), the literary future (6c) and the perfect subjunctive (6d) where the clitic host is typically consonant-initial, but in any case not necessarily vowel-initial. (6)

a.

Dorința dedesire.the of spune(-o) say.INF(=it) “The desire to say it”

a to

(o) it=

30

Chapter Two

b.

c.

d.

Pentru a (o) fi spus(-o) for to it= be said(=it) “In order to have said it” (O) vom spune(-o) it= we.will say.INF(=it) “We will say it” Nu mă aștept să (o) not CL.REFL.1SG= I.wait that.SBJV it= fi rezolvat(-o) be solved(=it) “I don’t expect myself/you/him/her/us/them to have solved it”

There can be little doubt that phonological approaches offer absolutely no explanation for such cases which, as we shall see below, must instead be explained in syntactic terms.

2.2 Diachronic evidence Equally from a diachronic perspective, the proposed purely phonological accounts prove misplaced. Firstly, the apparently illicit phonological sequences observed in (1b-d) prove entirely grammatical in older stages of the language (de Kok 1989; Croitor 2012: 124; Dragomirescu 2013: 194; Vasilescu 2013: 387): (7)

a.

b.

c.

o-au întors (ORom.) it=have.3 turned “they have turned it around” numai în Africa o ar fi only in Africa her= would.3 be descoperit Dumnezeu (ORom.) discovered God “God would only have discovered her in Africa” sau va adăoga-o sau o or he.will increase.INF=it or it= va împuțina (ORom.) he.will reduce.INF “he will either increase it or reduce it”

In contrast to (1b-d), we see, for instance, that in (7a-b) the feminine singular accusative clitic occurs in proclisis to the present perfective and

Parallels in Clausal and Nominal Structures: Romanian Clitic Placement

31

conditional auxiliaries au and ar, respectively. Also of interest is example (7c) where, even in the presence of the consonant-initial future auxiliary va “he will”, the feminine singular accusative clitic is shown to occur both in enclisis to the infinitive and in proclisis to the auxiliary within the same sentence. Such facts clearly highlight how the placement of the clitic, at least in this case, is not sensitive to phonological factors. Examples such as (7a-b) with proclisis of the clitic in conjunction with a vowel-initial auxiliary are not rare in old Romanian but, as Stan (2013: 170) notes, they prove particularly frequent from the sixteenth to the eighteenth centuries and are especially common in learnèd texts (cf. 8a-c). (8)

a.

b.

c.

o a pus (GB.115v) it= he.has put “he has placed it” o am vândut it= I.have sold “I have sold it” (DÎ, LX, zapis, Galați, 1577) o au făcut (CIst.9v) it= have.3 done “they have done it”

Indeed, Densusianu ([1901]1997: 727) goes even further and claims that “[t]oujours contrairement à l’ordre courant d’aujourd’hui, o est placé d’habitude avant la forme périphrastique du parfait” [underlining A.L.] and provides examples such as (9a-c): (9)

a.

b.

c.

o am aflat (CC2.9) it= I.have found “I found it” mi-o au trimes (CC2.9) me=it have.3 sent “they sent it to me” o amu săpatu (Măn.Taz.5) it= we.have dug “we have dug it up”

This observation highlights how the distribution of the clitic o witnessed for modern Romanian which, we have seen, has been argued, at least in part, to be driven by phonological considerations represents a relatively recent development, inasmuch as the feminine singular accusative clitic regularly appeared before vowel-initial auxiliaries.

32

Chapter Two

However, a more nuanced picture of the distribution of o in old Romanian is presented in Nicolae and Niculescu (2016:§2.2.3.7) who note that the clitic is either proclitic (cf. 10a) or enclitic (cf. 10b) in old Romanian with a number of verb forms (namely, present, compound past, voi-infinitival future and perfect future, analytic pluperfect, conditional, perfect subjunctive, and gerundial periphrases). (10)

a.

b.

de o au făcut vin (CC1.4v) and it= he.has made wine “and he turned it into wine” dac -ar fi dat-o (A.63r) if would.3 be given=it “if he had given it”

Below in Table 2 we reproduce the available figures provided by Nicolae and Niculescu (2016) in relation to the distribution of the feminine singular accusative clitic in old Romanian texts which, for the purposes of our presentation, can be split into two distinct diachronic periods, the first ranging from 1600 until 1650 and the second from 1650 until 1780.

ITM 1651–1750 DPV.1673 CÎ.1678 Mârg. 1691 VS.post1700 CIst.1700–50 CLM.1700–50 NL.~1750–66 ITM 1751–95 Prav. 1780 Total

ITM.1600−50 DRH.A.XXIII 1635–36 DRH.B.XXXI 1646 NT.1648 Total

Con LitFut First old Romanian Period (1600–50) 52/48% — 89/11% 100/0% 100/0% 78/22% (6 occ.) 100/0% 97/3% 100/0% 79/21% 100/0% 100/0% Second old Romanian Period (1651-1780) 48/52% — 100/0% — — 100/0% — 100/0% P(/E) 100/0% — 100/0% — 19/81% — 17/83% 36/64% 15/85% 30/70% — — P(/E) 63/37% 57/43% 30/70% 75/25%

PrsPrf

— — — — — — — — — — —

100/0%

100/0% — — — — — — — — — — —

100/0% (1 occ.)

GerFut

100/0% (2 occ.)

PluPrf

Table 2: Distribution of proclisis/enclisis of o in old Romanian in Nicolae and Niculescu (2016:§ 2.2.3.7)

Parallels in Clausal and Nominal Structures: Romanian Clitic Placement

33

34

Chapter Two

As Table 2 clearly illustrates, a comparison of texts from both periods reveals some significant changes in the placement of the feminine singular accusative clitic through time. In particular, we witness how in the earlier period proclisis of o undoubtedly predominates over enclisis: not only is proclisis the only order found with the consonant-initial auxiliary voi employed to form the literary and gerundial futures, but also with the vowel-initial conditional (e.g. aș) and pluperfect (e.g. am) auxiliaries. Moreover, proclisis easily predominates over enclisis with the vowelinitial present perfect auxiliary (e.g. am). By contrast, in the second period, and especially towards the latter end of that period, we see an unmistakable and significant increase in the distribution of enclisis to the detriment of proclisis. In the literary future formed with the auxiliary voi, enclisis finally emerges either as a serious competitor to proclisis (e.g. Prav.1780) or as the predominant pattern (e.g. NL.~1750–66), reflecting in large part the situation noted above for the modern language where proclisis and enclisis appear to alternate freely. We also witness a change in conjunction with the vowel-initial auxiliaries where enclisis emerges as an option, albeit generally still less common than proclisis, with the conditional auxiliary (e.g. aș) and, above all, with the present perfect auxiliary (e.g. am) where it soon comes to prevail over proclisis. Overall, then, the most salient diachronic development to emerge from Table 2 is the gradual increase of enclisis of o especially before vowelinitial auxiliaries, largely heralding the modern distribution witnessed in (1b-f). Crucially, however, the evidence of Table 2 highlights that the gradual emergence of enclisis is not ultimately driven by phonological factors but, rather, affects both vowel-initial and consonant-initial auxiliaries alike.

2.3 Diatopic evidence From a diatopic perspective, the purely phonological accounts prove once again highly problematic. More specifically, there are a number of regional varieties of the language in which the restrictions on proclisis witnessed in (1b-g) for the standard language do not hold. For instance, in his Tratat de dialectologie românească Rusu (1984: 303, 343, 381) highlights large areas of northeastern, northern and central Romania in which the feminine singular accusative clitic continues today to occur proclitic to the vowel-initial perfective auxiliary (e.g. am) on a par with what was observed in §2.2 for previous stages of the language. Nicolae and Niculescu (2016: 61) point towards a similar picture, highlighting how “[p]reverbal o with the compound past is still attested

Parallels in Clausal and Nominal Structures: Romanian Clitic Placement

35

dialectally, in Crişana, Maramureş, Transylvania”. Broadly speaking, the general phonological properties of all these modern varieties are largely similar, if not identical in numerous cases, to those attested in the standard more southerly varieties. Nonetheless, proclisis of the clitic o to the same vowel-initial auxiliaries that today favour enclisis in the standard language proves entirely grammatical in these dialectal varieties, strongly suggesting that the distributions of proclisis and enclisis in both the standard and the regional varieties is not driven by phonological considerations.

2.4 Interim conclusion The preceding discussion has incontrovertibly demonstrated how from a variety of synchronic, diachronic and diatopic perspectives purely phonological explanations of the distribution of o in modern Romanian exemplified in (1b-g) are ultimately untenable. Although acknowledging some growing correlation between the occurrence of enclisis and vowelinitial auxiliaries in the modern language, the core facts simply cannot be accommodated under phonological accounts inasmuch as (i) the supposed illicit vocalic sequence produced by proclisis (cf. 1b-d) is not phonologically ill-formed at any point in the history of the language nor in any of its regional varieties; and (ii) enclisis is also frequently, and increasingly so through time, attested in contexts where the supposed illicit vocalic sequence simply does not arise under proclisis, as with the (perfective) infinitive cf. (1e), the literary future (cf. 1f) and the perfective subjunctive (1g). It follows that the distribution of the clitic o must primarily be understood in structural terms, as will be argued in §4 below. Nonetheless, we cannot deny that the structural principles underlying the distribution of the feminine singular accusative clitic in the modern language also give rise to a surface distribution which, synchronically, is increasingly susceptible to partial reanalysis along phonological lines. Consequently, in attempting to account for the distribution of the clitic o today some reference to phonological information is ultimately necessary to make sense of the relevant facts, as observed by Dobrovie-Sorin (1994: 22, 73) and Rîpeanu, Tasmowski and Vasilescu (2013: 251f.). Therefore statements like those in (11), although facing many of the problems raised in relation to the quotes in (3a-c) above, have to be borne in mind. (11)

a.

“if the auxiliary has an initial vowel, it cannot host the clitic o […] if the auxiliary has an initial consonant, the clitic o can be either hosted by V1 […], or by V2 […]” (Dragomirescu 2013: 193)

36

Chapter Two

b.

c.

“This difference between the future and the conditional paradigms is due to phonological rules: the future auxiliary presents an initial consonant, while the conditional auxiliary starts with a vowel.” (DobrovieSorin 1994: 22 n. 31) “for the feminine accusative singular clitic o, that is its position in those clusters that contain an auxiliary beginning with a vowel: am văzut-o ‘I saw her’, aș vedea-o ‘I would see her’, *o am văzut, *o aș vedea. The postverbal position of the clitic o must at least partially be due to some morpho-phonological constraint which prohibits the adjacency of two vowels: *o oi, *o aș, *o am.” (Rîpeanu Reinheimer, Tasmowski and Vasilescu 2013: 251; [underlining A.L.])

3. Clitic positions Much recent research into the syntax of Romance cliticization (see, among others, Parry 1994; 1995; 2005; Tortora 2002; 2015; Ledgeway and Lombardi 2005; Benincà and Tortora 2009; 2010; Benincà and Pescarini 2016; Roberts 2016) has revealed that object clitics can target as many as three distinct positions within the clause, roughly corresponding to the vP, TP and CP domains as sketched in (12). (12)

[CP

Cl3

[TP

Cl2

[vP

Cl1

]]]

For our current purposes, we ignore here the highest clitic site (Cl3) situated in the C-domain (see, though, Pescarini this volume for Romance evidence for a C-related clitic site), but focus instead on the lower TPrelated (Cl2) and vP-related (Cl1) sites which prove most relevant for the present discussion. Evidence for these two clitic sites is not difficult to come by in Romance. It will suffice here to consider just two examples from Italo-Romance. The first concerns Piedmontese dialects such as Borgomanerese which show generalized enclisis. As argued by Tortora (2002; 2010; 2014; 2015), generalized enclisis in Borgomanerese involves cliticization to the lower portion of the clause within the aspectual field, such that the verb, which raises to a higher position above the aspectual field, is often separated from such clitics by an interpolated aspectual adverb lexicalizing one of the specifiers of the intervening aspectual projections. This is unambiguously shown in (13a) where the terminative aspectual adverb piö “no longer”

Parallels in Clausal and Nominal Structures: Romanian Clitic Placement

37

intervenes between the raised finite verb vœnghi “I see” and the object clitic lla “her” which we take to lexicalize the lowest clitic site that we have labelled Cl1. In compound perfective periphrases, on the other hand, such object clitics raise higher in the clause to left-adjoin to the perfective auxiliary but, in contrast to most other Romance varieties, typically display in such cases a second copy of the same pronoun enclitic to the participle, witness the example in (13b). (13)

a.

I

vœnghi piö-lla I.see no.more=her “I no longer see her” SCL.1SG

b.

(Borgomanerese; = Cl1) L Piero l’ à sempri the Piero it= has always mangià-llu eaten=it “Piero has always eaten it” (Borgomanerese; = Cl2…Cl1)

Putting aside technical details, we take examples such as (13b) to involve the multiple PF spell-out of the pronoun which is lexicalized both in a lower vP-related position on the active participle and in a higher TPrelated position proclitic to the auxiliary. We therefore take structures such as (13b) to provide direct evidence for the lower and higher clausal clitic sites Cl1 and Cl2 hypothesized in (12) above. Our second example concerns the eastern Abruzzese dialect of San Valentino recently studied by Benincà and Pescarini (2016). Among other things, the authors show that clitic placement in Sanvalentinese is characterized by a degree of apparent optionality. For example, in perfective constructions object clitics may appear both in proclisis (14a) and enclisis (14b) to the finite auxiliary or, alternatively, in proclisis (15a) and enclisis (15b) to the participle. (14)

a.

mǝ me=

b.

'ajǝ mǝ l I.have =me =it “I”ve eaten it up”

l it=

'ajǝ I.have

mǝɲ'ɲɐtǝ eaten (San Valentino; = Cl2) mǝɲ'ɲɐtǝ eaten (San Valentino; = Cl2)

38

(15)

Chapter Two

a.

b.

'ajǝ I.have

dʤa mǝ already me=

lu mǝɲ'ɲɐtǝ it= eaten (San Valentino; = Cl1) 'ajǝ dʤa mǝɲ'ɲɐtǝmǝlu I.have already eaten.=me=it “I’ve eaten it up” (San Valentino; = Cl1)

Convincingly, the authors show that such proclisis-enclisis alternations represent the surface output of variable verb movement, with enclisis obtaining whenever the finite (cf. 14b) or participial (cf. 15b) verb raises over the projection hosting the object clitics witnessed under proclisis in (14a) and (15a). Abstracting away from the proclisis-enclisis alternation in (14)-(15), what these examples highlight for our present purposes is the availability once again of two clitic placement sites, one located in the Tdomain (Cl2) hosting clitics in conjunction with the perfective auxiliary and the other located in the v-domain (Cl1) hosting clitics in conjunction with the active participle. Italo-Romance examples like those from Borgomanerese and San Valentino highlight therefore the presence of two clitic sites within the sentential core. However, various diachronic and diatopic varieties of Romanian also offer similar evidence for these Cl2 and Cl1 sites in clitic duplication structures where both positions appear to be simultaneously lexicalized. For example, in old Romanian (16th—19th centuries) it is not uncommon in periphrastic constructions to find object clitics realized both on the auxiliary and the associated non-finite lexical verb (Densușianu [1901]1997: 707; Dobrovie-Sorin 1994: 75; Dragomirescu 2013: 194; Vasilescu 2013: 387), a usage which proves particularly frequent in texts from central and northern Romania, witness the quotes in (16a-b): (16)

a.

b.

“În textele de la începutul secolului se întâlnește destul de frecvent fenomenul repetării cliticului, acesta fiind utilizat înainte și după forma verbală compusă. Atestările sunt mai ales din Transilvania” (Croitor 2012: 124) “O particularitate regională nordică, prezentă, mai ales, în texte literare moldovenești și transilvănene din secolele al XVI-lea-al XVIII-lea, este exprimarea redundantă a cliticului înainte și după verb” (Stan 2013: 170)

Parallels in Clausal and Nominal Structures: Romanian Clitic Placement

39

Below we reproduce some of the relevant examples which show that object clitics of all types, and not just o, are found in duplication structures involving both the perfect indicative (17a-d), future (18a-c), conditional (19) and perfect subjunctive (20) periphrases. (17)

a.

b.

c.

d.

(18)

a.

b.

c.

(19)

o it=

(20)

[…]

te-am mâhnitu-te (...) te-am you=I.have saddened=you you=I.have scosu-te (...) te-am izbăvitu-te removed=you you=I.have saved=you “I have saddened you…I have removed you…I have saved you” (DPar.69v) o au dat-o (...) o au it= have.3 given=it it= have.3 slobozit-o (...) o au tins-o released=it it= have.3 kept=it “they have given it…they have released it…they have kept it” (ULM 99) l-ai tocmitu-l it=you.have negotiated=it “you have negotiated it” (Mol.1725, in ILRL: 352) și scrisoarea o am and letter.the it= I.have făcut-o done=it “and I have written the letter” (ARG) te voi chema-te (DVS.90r) you= I.will cal.INF=you “I will call you” m-oi lăuda-mă (DVS.89v, 90r) me=you.will praise.INF=me “you will praise me” o va cerca-o (Filerot.124) it he.will examine.INF=it “he will examine it” aș fi uitat-o (Filerot.141–2) I.would be forgotten=it “I would have forgotten it” să nu o fi știut-o that.SBJV not it= be known=it “[…] that I didn’t know it” (Maior)

40

Chapter Two

It would be incorrect to interpret such clitic duplication examples as rare or exceptional aberrations from the norm, inasmuch as they are widely attested in the history of Romanian. Indeed, Nicolae and Niculescu (2016:§2.2.7) observe that, although the phenomenon is rare in the early sixteenth century, from 1640 onwards such clitic duplications become very frequent in the analytic finite periphrases expressing the compound past, the future and the conditional. Structures like those in (17)-(20) are not however limited to earlier periods of the language, but continue to the present day in a number of regional varieties of Romanian, especially those spoken in the centre and north of the country, as the observations in (21) highlight.1 (21)

a.

b.

c.

“Dans la langue populaire, surtout en Moldavie, il est assez fréquent qu’un pronom atone devant le verbe soit répété immédiatement derrière le même verbe” (Sandfeld and Olsen 1936: 102) “phenomena that used to exist in older (see also dialectal) [underlining A.L.] varieties of Romanian” (Dobrovie-Sorin 1994: 75) “The pronominal clitic can be doubly realized […] it is also attested in present-day Romanian dialects of Banat, Maramureş, Transylvania, and Bucovina” (Nicolae and Niculescu 2016: 61)

Below in (22a-e) we illustrate a number of representative examples from the modern period. (22)

a.

b.

1

eu v’ oiu primi-vă (Vas.Pov.17) I you= I.will receive.INF=you “I will welcome you Ț’ am spusu-ți, frate Alexandru, you= I.have told=you brother Alexandru că (Vas.Pov.161) that “Brother Alexandru, I’ve told you that […]”

A simple search of the internet also produces a large number of examples of clitic duplications structures, particularly in conjunction with the feminine singular accusative clitic o in the perfect subjunctive and, to a lesser extent, the literary future (Irina Nicula, p.c.), suggesting that the phenomenon continues below the radar of the standard language in the informal registers of many speakers even beyond the central and northern regions of the country.

Parallels in Clausal and Nominal Structures: Romanian Clitic Placement

c.

d.

e.

41

nici nu v’ oiu judeca-vă neither not you= I.will judge.INF=you eu (Vas.Pov.162) I “neither will I judge you” slujba cu care m’ai service.the with which me=you.have însărcinat o voiu împlini-o charged it= I.will execute.INF=it “the job which you’ve charged me with, I’ll do it” (Isp. 36) L-am văzutu-l / o voi him=I.have seen=him it= I.will lua-o / l-a spălatu-l/ take.INF=it it=he.has washed=it m-oi bucura-mă me=I.will rejoice=me “I have seen him / I will take it / he has washed it / I will be happy” (Bredemeier 1976: 38f.)

Analogously to the Italo-Romance examples examined in (13)-(15), the examples taken from different diachronic and diatopic Romanian varieties observed (17)-(20) and (22) provide us with incontrovertible evidence of at least two clitic sites within the sentential core which we have associated with the TP- and vP-domains (viz. Cl2…Cl1). Quite simply, then, we propose to interpret the Romanian clitic duplication structures as a case of multiple spell-out of the relevant clitic at PF in Cl1 and Cl2 as it raises through the clause, such that high V(P)-movement of the non-finite verb (cf. Cornilescu and Nicolae 2013; Schifano 2014; 2015a,b; Nicolae 2015) over Cl1 gives rise to phonological encliticization of the lowest copy to the participle or infinitive at PF. This is sketched informally in (23): (23)

[TP/InflP Cl2 Aux PtP/Inf [vP Cl1

PtP/Inf…[VP PtP/Inf Cl0]]]

Adopting the structural representation in (23), we assume that in modern Romanian enclisis of the feminine singular accusative o in contrast to the proclisis of all other clitics in the relevant periphrases can be derived as follows: the non-finite participial or infinitival verb undergoes high Vmovement to a position within the T-domain whereas clitics, first-merged as DPs within the lexical VP (viz. Cl0), raise out of the VP first transiting

42

Chapter Two

through the lower vP-related clitic position (Cl1) before surfacing in the higher TP-related clitic position (Cl2). In the standard case, therefore, the object pronoun procliticizes to the auxiliary (or finite V) first-merged in (or raised to) the T-domain at PF (24). In the exceptional case of the feminine singular accusative clitic o, by contrast, the higher clitic position is simply not available and the clitic surfaces in the lower vP-related position from where it encliticizes at PF to the participle or infinitive immediately to its left (25). (24)

(25)

[TP Cl2 Aux PtP/Inf [vP Cl1 PtP/Inf… [VP PtP/Inf Cl0]]] (*o) am oprit/opri -o her=we’ve/’d stopped/stop.INF =her “We have stopped her/We would stop her” [TP Cl2 Aux PtP/Inf [vP Cl1 PtP/Inf… L-am oprit/opri (*-(u)l) him=we’ve/’d stopped/stop.INF =him [VP PtP/Inf Cl0]]] “We have stopped him/We would stop him”

In the next section we shall explore the reasons why in the modern language the feminine singular accusative clitic is banned from raising to the higher clitic position in such periphrases but, rather, is forced to surface in the lower clitic position.

4. Syntactic account As already observed, previous phonological analyses claim that “the contrast in (48) vs (49)a-b [cf. 24 vs 25] cannot be assumed to be syntactic: rules of syntax are probably not sensitive to specifications such as ‘feminine singular’.” (Dobrovie-Sorin 1994: 22). Although agreeing that syntax is probably not sensitive to the “feminine singular” feature, we shall argue in what follows that the contrast between (24) and (25) should indeed be handled in syntactic terms. In particular, we claim that the formal identity of the feminine singular accusative clitic (ILLAM >) o and the feminine singular nominative-accusative indefinite article (UNAM >) o should be taken seriously, inasmuch as it is responsible for a parallel between clausal and nominal syntax that underlies the exceptional behaviour of the placement of the object clitic o. For instance, we see in (26a-b) that, synchronically, the morpheme o may lexicalize either a Drelated position in the nominal domain where it surfaces as the indefinite article, or a D-related position within the clausal domain where it spells out

Parallels in Clausal and Nominal Structures: Romanian Clitic Placement

43

the phi-features of direct object on the finite verb. In short, we see that irrespective of its distribution within the nominal or clausal domains, o invariably functions as a D-related element. (26)

a.

b.

Ionuț cumpără o Ionuț buys a.F.SG frumoasă beautiful.FSG “Ionuț buys a beautiful house” Ionuț o cumpără Ionuț it.FSG= he.buys “Ionuț buys it”

casă house.FSG

Capitalizing on parallels between nominal and clausal structure,2 we can now make sense of the Romanian facts. Following Grimshaw (2000), we adopt general assumptions about the structure of the sentential core before the left periphery (LP) which we take to be composed of a T(ense)domain and a v(erbal)-domain both of which, as we have already seen, come with their own clitic positions, as illustrated informally in (27). (27)

[LP … [TP Cl2-T… [vP Cl1-v… [VP …V]]]]

In a similar fashion, it is generally assumed (cf. Cinque 1995; Lyons 1999: 298–303; Longobardi 2001:§3.3) that, above the lexical domain (viz. NP), the functional structure of nominal expressions can also be divided into two hierarchical domains labelled DP and nP (cf. 28a). While the former is typically headed by the definite article, the head of nP is the position variously lexicalized by the indefinite article or cardinal numbers (hence often labelled CardP or NumP in some studies). These assumptions map very neatly onto Cornilescu’s (1992) study of the Romanian determiner system where she makes similar structural distinctions, albeit with the difference that she assumes a recursive DP structure in which the higher DP hosts the definite article (licensing definiteness and Case) and the lower one indefiniteness and the indefinite article (28b). Crucially, we see that under both analyses nominal structures are taken to contain two Drelated positions. (28)

a. b.

[LP [LP

[DPthe-D [DP-ul-D

[nPa-n [NP N]]]] [DPun-D [NP …N]]]]

2 Cf. Abney (1987), Szabolcsi (1994), Lyons (1999), Giusti (2005; 2006; 2014), Bošković (2010), Ledgeway (2015).

44

Chapter Two

If we now put together the clausal and nominal structures sketched in (27) and (28), we immediately observe a striking parallelism between the two structures, inasmuch as both display two D positions above the lexical domain: (29)

a.

[LP

[TP Cl2-T

b.

[LP

[DP the-D

[vP Cl1-v [VP …V]]]] (clausal structure) [nP a-n [NP …N]]]] (nominal structure)

Returning to Romanian clausal and nominal structures, we can now see how the D positions observed in (29a-b) are lexicalized in Romanian: (30)

a.

b.

[LP … [TP Cl2-T… [vP Cl1-v… [VP …V]]]] o opresc (= “her= I.stop”) am oprit -o (=“I’ve stopped =her”) [LP … [DP the-D… [nP a-n… [NP …N]]]] oala (= “pot.the”) o oală (= “a pot”)

Quite simply, then, our proposal to explain the enclisis of modern Romanian o appeals to the idea that the (multiple) spell-out of Cl1 observed in clitic duplication structures such as (17)-(20) has been exceptionally retained in conjunction with the feminine singular accusative clitic pronoun which continues to spell-out the lower D position. This particular behaviour of o qua pronominal clitic is to be explained as a result of the increasing analogical influence of o qua indefinite article which, we have seen, lexicalizes the lower D position within the nominal domain. In particular, we can increasingly think of o as a single lexical item, namely a D head which is un(der)specified for definiteness, namely [D, FSG, -GEN/DAT]. We thus see the convergence of the syntax of o qua pronominal clitic and o qua indefinite article in favour of the generalization of a single placement rule which increasingly associates o in both clausal and nominal domains with the lower of the two D positions. On this view, the exceptional enclisis of pronominal o in various aspectual, temporal and modal periphrases, which we take to involve the spell-out of the lower D position, is subsequently open to reinterpretation as a result of its phonological shape, in that most, though not all, of the potential clitic hosts involve vowel-initial auxiliaries. Consequently, the surface distribution of o, as revealed by the emerging patterns in earlier stages of the language witnessed in Table 2, can today be partially

Parallels in Clausal and Nominal Structures: Romanian Clitic Placement

45

reanalyzed synchronically as phonologically-driven in that enclisis proves most robust in conjunction with vowel-initial auxiliaries (cf. Table 1). This analysis based on the parallel behaviour of the feminine singular D-element o in both clausal and nominal domains explains a number of otherwise puzzling facts and, at the same time, makes a number of significant predictions. First, the analysis sketched above naturally explains why o alone surfaces in the lower vP-related clitic position, while all other pronominal clitics surface in the higher TP-related clitic position: only the feminine singular accusative clitic o coincides formally with the indefinite article, which we know targets a lower D position, whereas all other pronominal forms are morphophonologically distinct from the forms of the indefinite article. For example, the masculine singular accusative clitic (î)l and the masculine singular indefinite article un are formally quite distinct and can hardly be interpreted as one and the same element. Indeed, the masculine singular accusative clitic (î)l is, if anything, superficially similar (and diachronically related) to at least one of the allomorphs of the masculine singular nominative-accusative definite article -(u)l which, we have seen (cf. 28a-b), targets the higher of the two D positions. In short, the only pronominal object clitic which could have been analogically influenced by the indefinite article and hence increasingly associated with the lower D position is the feminine singular accusative clitic o which coincides entirely with the feminine singular nominative-accusative indefinite article, whereas such an analogical model was simply not available with all other pronominal clitics. Second, the claim that the lower placement of the pronominal clitic o is ultimately a consequence of its formal homophony with the indefinite article is indirectly supported by a comparison with other Daco-Romance varieties. For example, in Aromanian the feminine singular accusative clitic pronoun and the feminine singular indefinite article are morphophonologically quite distinct, namely u(â)/o and unâ, respectively. As a consequence, the two items remain distinct in Aromanian and the lower placement of the indefinite article unâ in the nominal domain has not influenced the placement of the feminine clitic pronoun u in the clausal domain. Thus we see in (31a-b) that, in contrast to Romanian, the feminine singular accusative clitic pronoun u in the Aromanian perfective periphrasis does not encliticize to the participle (viz. Cl1) but, rather, procliticizes to the vowel-initial finite auxiliary (viz. Cl2) on par with all other object clitics (cf. Capidan 2005: 537):

46

(31)

Chapter Two

a.

b.

Maria uavemu mârtatâ Maria her= we.have married “As for Maria, we married her off” (Aro., Caragiu Marioțeanu and Saramandu 2005) ațeá ńíca uaveá loatâ that.one.F small.F her= he.had taken “the small one, he had taken her (for his wife)” (Aro., Nevaci 2006: 113)

Third, we can now also explain why, despite the phonological context, in the modern language o may—optionally or substandardly—occur enclitic to the non-finite infinitive or participle even in conjunction with consonant-initial (auxiliary) verbs (cf. 1e-g). In particular, encliticization frequently arises in this context since un(der)specified o is increasingly associated with the lower D position and hence surfaces in Cl1 irrespective of the phonological shape of the main (auxiliary) verb in the T-domain. Fourth, by the same token we may now also explain why pronominal o may optionally continue to occur in proclitic position with consonantinitial (auxiliary) verbs (cf. 1e-g). Specifically, we have identified an ongoing convergence involving both pronominal and article instantiations of o as an un(der)specified D favouring the lower D position, but this ongoing process of reanalysis is not yet complete such that the clitic may still lexicalize the higher D position in contexts where the phonological environment originally sanctioned it (cf. Table 2). Finally, we must turn to the question of why enclisis of pronominal o is impossible with synthetic finite lexical verbs, including those that are vowel-initial (cf. 1a). Quite simply, we assume following Cinque (1999) that the T-domain is made up of various functional projections that can be broadly subsumed within three fields associated with Mood, Tense and Aspect, as sketched in (32). Assuming furthermore that, in the absence of an auxiliary, finite lexical verbs raise to the highest available functional head in Romanian, namely a Mood head (cf. Nicolae 2015:ch. 5; Schifano 2015a,b), the clitic o must target the higher clitic site Cl2, from where it procliticizes onto the finite lexical verb, since it if were to surface in the lower clitic site Cl1 it would not be able to encliticize onto the finite verb which has raised to the highest T-related position (cf. 32a). In auxiliary structures, by contrast, the highest T-position is lexicalized by the auxiliary, such that the non-finite verb is forced to raise to a lower position

Parallels in Clausal and Nominal Structures: Romanian Clitic Placement

47

from where it can act as an appropriate enclitic host for the clitic o whenever it surfaces in the lower Cl1 position (cf. 32b).3 (32)

[TP Cl2-Mood

a.

b.

T [vP Cl1-v-Asp

[VP V

o opresc opresc (*-o) opresc her=I.stop “I stop her” am oprit-o oprit I.have stopped=her “I have stopped her”

DP-Obj ]]]

opresc o

oprit

o

5. Conclusions To conclude, we have argued that, on a par with many other Romance varieties, Romanian provides evidence for at least two clitic positions within the sentential core associated with the TP- and vP-domains, namely Cl2…Cl1. In the past, and still today in some central and northern regional varieties, both positions are often simultaneously spelt out at PF with copies of the same pronoun being pronounced in both the lower and higher clitic positions. In contrast to all other clitic pronouns, in the standard language the lower clitic position is today only exceptionally lexicalized by the feminine singular accusative clitic o (< ILLAM) on account of its being formaly homophonous with the indefinite article o (< UNAM). In particular, we have seen how within nominal structures the indefinite article is standardly taken to lexicalize the lower of two D positions (= nP), such that through time there is an analogical tendency for the homophonous pronominal clitic o to also target the lower D position (viz. Cl1 in vP) in the corresponding clausal structure. As a result, the two occurrences of o, though historically from different sources, are increasingly interpreted as a single functional item, namely an 3

Depending on how one defines the phase boundary in a cartographic approach given that vP presumably includes a number of Cinque’s (1999) lower aspectual and possibly temporal projections, these facts should ultimately be derived from the Phase Impenetrability Condition (PIC). For example, if we take the finite lexical verb in (32a) to surface in the highest CP phase, then it will be not be sent to PF in the same cycle as the clitic o in Cl1 contained in the lower vP-related phase. In (32b), by contrast, we can take the lower position lexicalized by the participle to correspond to a position within the lower vP-related phase which enters PF in the same cycle as the clitic in Cl1. Clearly, more work is needed to explore the detailed consequences of phase theory and its potential compatibiliy with cartographic approaches.

48

Chapter Two

un(der)specified D which targets the lowest available D position within both the nominal and clausal domains. Finally, given that enclisis of pronominal o predominantly emerges in periphrases in which the auxiliary is vowel-initial, its distribution has been partially reinterpreted synchronically as a surface effect of a language-particular phonological constraint which rules out sequences, and hence proclisis of o, in which two non-front vowels are immediately contiguous. This explains why many previous analyses have tried, without success, to explain the relevant facts purely in phonological terms, although, as we have seen, the diachronic and synchronic distribution of enclisis of o must ultimately make reference first and foremost to structural considerations.

Corpus A.1620 = Alexandria. Ed. F. Zgraon, Bucharest: Fundaţia Naţională pentru Ştiinţă şi Artă, 2005 (Cele mai vechi cărţi populare în literatura română, 11). (South Transylvania, Braşov or Haţeg) CC1.1567 = Coresi, Tâlcul Evangheliilor. Ed.: Coresi, Tâlcul evangheliilor și molitvenic românesc, ed. V. Drimba, Bucharest, Editura Academiei Române, 1998, 31–187. (Transylvania, Wallachian subdialect; Gheție and Mareș 2001: 115) CC2.1581 = Coresi, Evanghelie cu învăţătură. Ed. S. Puşcariu, Al. Procopovici: Diaconul Coresi, Carte cu învăţătură (1581), vol. I, Textul, Bucharest, Socec, 1914. (Braşov) CIst.1700–50 = Constantin Cantacuzino, Istoria Ţării Româneşti. Ed.: Istoria Ţărâi Rumâneşti atribuită stolnicului Constantin Cantacuzino, ed. O. Dragomir, Bucharest, Editura Academiei Române, 2006, 145–202. (Bucharest) CÎ.1678 = Cheaia înţelesului. Ed. Ioannykij Haleatovskyi, Cheia înţelesului, ed. R. Popescu, Bucharest, Libra, 2000, 13–194. (Bucharest) CLM.1700–50 = Miron Costin, Letopiseţul Ţărâi Moldovei. Ed.: M. Costin, Opere, ed. P. P. Panaitescu, Bucharest, Editura de Stat pentru Literatură și Artă, 1958, 41–201. (Moldova) DÎ = Documente şi însemnări româneşti din secolul al XVI-lea, text stabilit şi indice de Gh. Chivu, M. Georgescu, M. Ioniţă, Al. Mareş, Al. Roman-Moraru, Bucharest, Editura Academiei Române, 1979. DPar.1683 = Dosoftei, Parimiile preste an, 1683, ed. M. Ungureanu, Iași, Editura Universității „Al. I. Cuza”, 2012, 95–356. (Moldova, Iași) DPV.1673 = Dosoftei, Psaltirea în versuri. Ed.: Dosoftei, Opere, 1, Versuri, ed. N. A. Ursu, Iași, Mitropolia Moldovei și a Sucevei, 1974, 3–1065. (Ukraine, Uniev) DRH.A = Documenta Romaniae Historica. A. Moldova, Bucharest, Editura Academiei Române, 1996: vol. XXIII (1635–1636); 2006: XXVIII (1645– 1646). (Moldova)

Parallels in Clausal and Nominal Structures: Romanian Clitic Placement

49

DRH.B = Documenta Romaniae Historica. B. Țara Românească, Bucharest, Editura Academiei Române, 1965: vol. XXI (1626); 1969: vol. XXIII (1630– 2); 1974: vol. XXIV (1633–4); 1998: vol. XXX (1645); 2003: vol. XXXI (1646); 2002: vol. XXXIV (1649); vol. XXXV (1650); 2006: vol. XXXVII (1652); 2009: vol. XXXVIII (1653). (Wallachia) DVS.1682–6 = Dosoftei, Viața și petreacerea svinților, Iași. (Moldova) Filerot = Istoria lui Filerot şi cu a Anthusei, [Braşov, c. 1770; ms. 1374, BAR, Bucureşti]; Angela Tarantino (ed.), 1996, Roma, Bagatto Libri. GB.XVI-XVII = Glosele Bogdan. Ed. M. Georgescu in I. Gheţie (coord.), Texte româneşti din secolul al XVI-lea, 422–38. (North Moldova) ITM = Însemnări pe de pe manuscrise şi cărţi vechi din Ţara Moldovei, ed. I. Caproşu and E. Chiaburu, Iași: Demiurg, 2008, vol. I (1429–1750), 130– 582; vol. II (1751–1795), 5–325. Măn.Taz. = Mănăstirea Tazlău, jud. Neamţ, 20 iunie 1598. Mărg.1691 = Mărgăritare. Ioan Gură de Aur, Mărgăritare, ed. R. Popescu, Bucharest, Libra, 2001, 11–493. (Wallachia, Bucharest) Mol = Molităvnic. Ed. A. Dumitran, A.-M. Gherman, A. Vanca, Alba Iulia, Editura Reîntregirea, 2009, 163–1075. (Alba Iulia) NL.~1750–66 = Ion Neculce, Letopiseţul. Ed.: Ion Neculce, Letopiseţul Ţării Moldovei şi O samă de cuvinte, ed. I. Iordan, Bucharest: Editura de Stat pentru Literatură și Artă, ed. a II-a, 1959, 31–388. (Moldova and Wallachia) NT.1648 = Noul Testament. Ed. Alba Iulia, Reîntregirea, 1998. (Alba Iulia) Prav.1780 = Pravilniceasca condică. 1780, ed. Colectivul pentru vechiul drept românesc condus de acad. A. Rădulescu, Bucharest, EdituraAcademiei, 1957 (Adunarea izvoarelor vechiului drept românesc scris, 2), 36–156. (Wallachia, Bucharest) ULM.~1725 = Grigore Ureche, Letopiseţul Ţării Moldovei. Ed. P. P. Panaitescu, Bucharest, Editura de Stat pentru Literatură și Artă, 1955, 57–210. (Wallachia, original from Moldova) VS.post1700 = Vedenia Sofianei. Ed. A. Timotin, E. Timotin, Bucharest, FundaţiaNaţională pentru Ştiinţă şi Artă, 2001 (Cele mai vechi cărţi populare în literatura română, 6), 133–49. (Oltenia, Râmnic)

References Abney, S., 1986, The English Noun Phrase in its Sentential Aspect, unpublished doctoral dissertation, MIT. Alboiu, G., V. Motapanyane, V., 2000, “The generative approach to Romanian grammar. An overview”, in: V. Motapanyane (ed.), Comparative Studies in Romanian Syntax, Amsterdam, Elsevier, 1–48. Benincà, P., D. Pescarini, 2016, “Clitic placement in the dialect of S. Valentino in Abruzzo citeriore”, Archivio glottologico italiano, 101, 37–65. Benincà, P., C. Tortora, 2009, “Towards a finer-grained theory of Italian participial clausal architecture”, University of Pennsylvania Working Papers in Linguistics, 15, 1, 17–26.

50

Chapter Two

Benincà, P., C. Tortora, 2010, “On clausal architecture: Evidence from complement clitic placement in Romance”’, in: V. Torrens, L. Escobar, A. Gavarró, J. Gutiérrez (eds), Movement and Clitics: Adult and Child Grammar, Newcastle, Cambridge Scholars Publishing, 219–237. Bošković, Ž., 2010, “On NPs and clauses”, unpublished manuscript, University of Connecticut. Bredemeier, J., 1976, Strukturbeschrankungen im Rumanischen, Tübingen, TBLVerlag. Capidan, T., 2005, Aromânii. Dialectul Aromân. Studiu lingivistic (Ediția a II-a), Bucharest, Editura Fundației Culturale Aromâne. Caragiu Marioțeanu, M., N. Saramandu, 2005, Manual de aromână. Carti trâ învițari armâneaști, Bucharest, Editura Academiei Române. Cinque, G., 1995, “On the evidence for partial N-movement in the Romance DP”, in: G. Cinque, J. Koster, J.-Y. Pollock, L. Rizzi, R. Zanuttini (eds), Paths towards Universal Grammar. Studies in Honor of Richard S. Kayne, Washington, Georgetown Uninversity Press, 85–110. Cinque, G., 1999, Adverbs and Functional Heads: A Cross-linguistic Perspective, New York, Oxford University Press. Cornilescu, A., 1992, “Remarks on the determiner system of Romanian: The demonstratives al and cel”, Probus 4, 189–260. Cornilescu, A., A. Nicolae, 2013, “Remarks on verb movement in modern Romanian’, presentation given at the Al XIII-lea Colocviu internațional al Departamentului de Lingvistică, București, 13–14 December. Croitor, B., 2012, “Pronumele”, in: G. Chivu, G. Pană Dindelegan, A. Dragomirescu, I. Nedelcu, I. Nicula (eds), Studii de istorie a limbii române. Morfosintaxa limbii literare în secolele al XIX-lea şi al XX-lea, Bucharest, Editura Academiei Române, 121–172. Densusianu, O., [1901] 1997, Histoire de la langue roumaine, Bucharest, Editura Grai și Suflet—Cultură Națională. Dobrovie-Sorin, C., 1994, The Syntax of Romanian, Berlin/New York, Mouton de Gruyter. Dragomirescu, A., 2013, “Complex predicates”, in: G. Pană Dindelegan (ed.), The Grammar of Romanian, Oxford, Oxford University Press, 191-201. Giusti, G., 2005, “At the left periphery of the Romanian Noun Phrase”, in: M. Coene, L. Tasmowski (eds), On Space and Time in Language, ClujNapoca, Clusium, 23–49. Giusti, G., 2006, “Parallels in clausal and nominal periphery”, in: M. Frascarelli (ed.), Phases of Interpretation, Berlin / New York, Mouton, 163–184. Giusti, G., 2014, “On Force vs. Case and Fin vs Num”, in: A. Cardinaletti, G. Cinque (eds), On Peripheries. Exploring Clause Initial and Clause Final Positions, Tokyo, Hituzi Syobo Publishing, 189–207. Grimshaw, J., 2000, “Locality and extended projections”, in P. Coopmans, M. Everaert, J. Grimshaw (eds.), Lexical Specification and Lexical Insertion, Amsterdam, Benjamins, 115–133.

Parallels in Clausal and Nominal Structures: Romanian Clitic Placement

51

Kok, A. de, 1989, “On the conjunct object pronoun systems of Rumanian and French: A comparative analysis”, International Journal of Rumanian Studies, 7, 29-49. Ledgeway, A., 2015, “Parallels in Romance nominal and clausal microvariation”, Revue roumaine de linguistique, LX, 2, 105–127. Ledgeway, A., A. Lombardi, 2005, “Verb movement, adverbs and clitic positions in Romance”, Probus, 17, 79–113. Legendre, G., 2000, “Optimal Romanian clitics: A cross-linguistic perspective”, in: V. Motapanyane (ed.), Comparative Studies in Romanian Syntax, Amsterdam, Elsevier, 227–264. Lombard, A., 1974, La langue roumaine. Une presentation, Paris, Klincksieck. Longobardi, G., 2001, “The structure of DPs: Some principles, parameters and problems”, in: M. Baltin, C. Collins (eds), The Handbook of Contemporary Syntactic Theory, Oxford, Blackwell, 562–603. Lyons, C., 1999, Definiteness, Cambridge, Cambridge University Press. Monachesi, P., 2005, The Verbal Complex in Romance. A Case Study in Grammatical Interfaces, Oxford, Oxford University Press. Nevaci, M., 2006, Verbul în aromană. Structură și valori, Bucharest, Editura Academiei Române. Nicolae, A., 2015, Ordinea constituenților în limba română: O perspectivă diacronică. Structura propoziției și deplasarea verbului, Bucharest, Editura Universității din București. Nicolae, A., D. Niculescu, 2016, “Pronominal clitics: Clitic ordering, clitic clusters”, in: G. Pană Dindelegan (ed.). The Grammar of Old Romanian, Oxford, Oxford University Press, 52–70. Parry, M., 1994, “Posizione dei clitici complemento nelle costruzioni verbali perifrastiche del piemontese”, in: G.P. Clivio, C. Pich (eds), At dël VIII Rëscontr antërnassional dë studi an sla lenga e la literatura piemontèisa, Alba, Famija albèisa, 247–259. Parry, M., 1995, “Some observations on the syntax of clitic pronouns in Piedmontese”, in: J.C. Smith, M. Maiden (eds), Linguistic Theory and the Romance Languages, Amsterdam / Philadelphia, Benjamins, 133–160. Parry, M., 2005, Parluma ‘d Coiri. Sociolinguistica e grammatica del dialetto di Cairo Montenotte, Savona, Editrice Liguria. Rîpeanu Reinheimer, S., L. Tasmowski, A. Vasilescu, 2013, “Pronouns (§§1.11.7)”, in: C. Dobrovie-Sorin, I. Giurgea (eds), A Reference Grammar of Romanian. The Noun Phrase, Amsterdam / Philadelphia, Benjamins, 231– 268. Roberts, I., 2016, “Object clitics” in: A. Ledgeway, M. Maiden (eds), The Oxford Handbook of Romance Languages, Oxford, Oxford University Press, 786– 781. Rusu, V. (ed.), 1984, Tratat de dialectologie românească, Craiova, Scrisul Românesc. Sandfeld, K., H. Olsen, 1936, Syntaxe roumaine. I. Emploi des mots à flexion, Copenhagen, Munksgaard.

52

Chapter Two

Schifano, N, 2014, “(Un)marked patterns of verb-movement: The case of Romanian”, in: R. Zafiu, A. Dragomirescu, A. Nicolae (eds), Diacronie și sincronie în studiul limbii române, Bucharest, Editura Universității din București, 191–201. Schifano, N., 2015a, “The paradigmatic instantiation of TAM. A novel approach to Romance verb-movement”, in: E. O. Aboh (ed.), Romance Languages and Linguistic Theory 2013. Selected Papers from “Going Romance” Amsterdam 2013, Amsterdam, John Benjamins, 83–102. Schifano, N., 2015b, Verb Movement: a Pan-Romance Investigation, unpublished doctoral thesis, University of Cambridge. Stan, C., 2013, O sintaxă diacronică a limbii române vechi, Bucharest, Editura Universității din București. Szabolcsi, A., 1994, “The noun phrase”, in: K. Kiss, F. Kiefer (eds), The Syntactic Structure of Hungarian, New York, Academic Press, 179–274. Tortora, C., 2002, “Romance enclisis, prepositions and aspect”, Natural Language and Linguistic Theory, 20, 725–757. Tortora, C., 2010, “Domains of clitic placement in finite and non-finite clauses: Evidence from a Piedmontese dialect”, in: R. D’Alessandro, A. Ledgeway, I. Roberts (eds), Syntactic Variation: The Dialects of Italy, Cambridge, Cambridge University Press, 135–149. Tortora, C., 2014, “Patterns of variation and diachronic change in Piedmontese object clitic syntax”, in: P. Benincà, A. Ledgeway, N. Vincent (eds), Diachrony and Dialects. Grammatical Change in the Dialects of Italy, Oxford, Oxford University Press, 218–240. Tortora, C., 2015, A Comparative Grammar of Borgomanerese, New York, Oxford University Press. Vasilescu, A., 2013, “Pronouns”, in: G. Pană Dindelegan (ed.), The Grammar of Romanian, Oxford, Oxford University Press, 379–409.

CHAPTER THREE SYNTACTIC VARIATION IN TWO SISTER LANGUAGES: A STUDY OF WORD ORDER IN OLD FRENCH AND OLD OCCITAN SAM WOLFE Christ Church, University of Oxford [email protected]

1. Introduction 1.1 Aims of the chapter In recent decades, a vast body of research has emerged within Romance linguistics, demonstrating clearly that rich patterns of morphosyntactic microvariation obtain across Romance speaking Europe. Whilst this is true of contemporary Romance varieties, the question of whether equally rich variation can be ascertained from the textual records available historically has been little explored. Based on a new database of medieval Romance texts, the present chapter sets out to establish whether microvariation is attested in the core clausal syntax of French and Occitan varieties of the 12th and 13th centuries. We will argue that, despite a number of significant points of continuity between these two sister languages, a number of ill-explored differences are attested which warrant further discussion due to their clear theoretical significance.

54

Chapter Three

1.2 Background 1.2.1 Old French and Old Occitan The notion that the Romance languages of the medieval period showed major morphosyntactic differences from their modern counterparts is not a novel one and dates back at least as far the 19th century (Tobler 1875, 1883, 1886; Thurneysen 1892; Mussafia 1888). Within a formal generative tradition, these differences have most prominently been accounted for through arguing that all or some of the medieval Romance languages were Verb Second (V2) languages._ We should note that this hypothesis has arguably been most fully developed and most widely accepted in the case of Old French (Adams 1987a, 1987b; Roberts 1993; Vance 1995, 1997; Labelle and Hirschbühler 2005; Mathieu 2006, 2009; Salvesen 2011, 2013), though V2 accounts of Old Occitan do appear in the literature (Vanelli, Renzi and Benincà 1986: 53f; Benincà 2004: 263-265, 2006: 63-64, 2013: 75; Vance, Donaldson and Steiner 2009: 318; Donaldson 2015: 163, 2016: 41). Restricting ourselves for the moment to French, we should note, from the outset, that the single most compelling argument in favour of a V2 analysis comes from so-called ‘Germanic inversion’ structures (Adams 1987b: 4; Roberts 1993: 56; Vance 1997: 78-79; Salvesen and Beck 2014: 223), such as that in (1), which are not licit in the modern language: (1)

Ja vos avoit il si longuement already you=have.3SG.PST he so long servi serve.PTCP “He has already served you such a long time” (Old French, Quête 119)

The V2 property is standardly analysed in the theoretically-informed literature as a structural requirement that finite verb movement target the vacant C(omplementiser) position, whilst phrasal movement of a maximal category targets a C-related specifier (see Cardinaletti and Roberts 2002: 153; Ledgeway 2008: 438 and Holmberg 2015 for similar definitions). The presence of the postverbal pronominal subject in (1), which occurs structurally lower than the finite auxiliary, is therefore revealing. The fact that the subject appears higher in the functional structure than the past participle servi ‘served’, standardly taken to demarcate the left boundary of the v-VP-complex (Cinque 1999 and many more), suggests it occupies Spec-TP. This analysis in turn leads us to conclude that the auxiliary avoit

Syntactic Variation in Two Sister Languages

55

‘has’ cannot be in T⁰ and must have undergone movement into the left periphery. Crucially, however, the V2 property in Old French is linked to a raft of other morphosyntactic properties beyond Germanic-inversion. In brief, these include the attestation of XPNon-Subject—VFin-(S(ubject)) orders to a far greater extent than is the case in the modern language (Vanelli, Renzi and Benincà 1986:§4.1; Adams 1987b: 4-5, 1988; Roberts 1993: 85-87; Hulk and Van Kemenade 1995: 235-236; Vance 1997: 43-47), as in the example with a fronted direct object in (2): (2)

Ceste avision vi li rois this vision see.3SG.PST the king Mordrains en son dormant Mordrain in his sleep “King Mordrain had this vision in his sleep” (Old French, Quête 135)

Sharp matrix/embedded asymmetries are also evident across the (later) Old French textual record, where the V2 syntax of matrix clauses gives way to an SVO order in non-root environments (Adams 1987b: 5; Vanelli, Renzi and Benincà 1987:§4.2; Roberts 1993: 142, 2007: 61-63; Jensen 1994: 359; Vance 1997: 133): (3)

si

manda li preudons a ask.3SG.PST the man to son frere qu’il li envoiast his brother that-he him.CL=send.3SG.PST.SBJV armes et cheval arms and horse “the man asked the brother to send arms and a horse” (Old French, Quête 117) SI

Although they are less directly relevant to our discussion in this chapter, Old French also patterns distinctly from its modern counterpart in a number of further grammatical properties. Most notably, this includes the presence of a Tobler-Mussafia clitic system (4,5),_ which is sensitive to the status of the clause-initial element(s) in the left periphery (Benincà 1995 et seq.) and an operation similar to Germanic Stylistic Fronting (5) (Maling 1990; Holmberg 2000), which may differ from the Germanic languages in certain respects (see Mathieu 2006, 2009 and Labelle 2007 for discussion):

56

(4)

(5)

(6)

Chapter Three

Vait s’en li pople go.3SG=REFL.CL=LOC.CL the people “The people goes away” (Old French, Alexis, cxxi. 1, Labelle and Hirschbühler 2005: 62) Ja la voloient en feu metre already her.CL=want.3PL in fire put.INF “They wanted to put her in the fire” (Old French, Cligès, 5936-7, Labelle and Hirschbühler 2005: 62) Qant levéi furent ti del when up be.PST.3PL from-the mangier eating “When they had finished eating” (Old French, Le Chevalier à la Charrette 1043, Mathieu 2006: 225)

There is a far smaller literature on Old French’s sister language, Occitan. There is nevertheless a small yet thorough body of descriptive work on Old Occitan grammar (Hamlin, Ricketts and Hathaway 1967; Jensen 1990, 1994; Kunert 2003) and Occitan data have figured in a number of theoretically-informed comparative studies of Medieval Romance word order (Vanelli, Renzi and Benincà 1986: 53f; Benincà 2004: 263-265, 2006: 63-64, 2013: 75; Sitaridou 2005: 366-369, 2012: 570-574). In several recent works focussing on the left periphery of Old Occitan, it has been argued to have a V2 syntax (Vance, Donaldson and Steiner 2009: 318; Donaldson 2015: 163, 2016: 41) with a richly articulated C-domain (Vance, Donaldson and Steiner 2009: 313-318; Donaldson 2016: 43, 47-50). However, it has been noted that Old Occitan patterns distinctly from Old French in licensing relatively widespread orders where the finite verb does not appear second in the linear ordering (Jensen 1994: 359-360; Kunert 2003:§3; Vance, Donaldson and Steiner 2009: 315; Sitaridou 2012: 570; Donaldson 2015: 9-10; 2016: 43-49). This intuition that Old Occitan is somehow distinct from its sister language as regards the V2 property is a significant one, which we seek to fully explore in the discussion that follows. The empirical goal of the chapter is therefore to better understand the precise areas of the core grammar in which Occitan patterns similarly to French and those areas where it shows subtle or larger-scale distinctions. Before continuing, we should also note that an exploration of word order in Old Occitan and Old French is of interest on other grounds, namely that the V2 hypothesis for both languages has become

Syntactic Variation in Two Sister Languages

57

controversial. Since Kaiser (2002), a number of doubts have been raised about the veracity of the claim that Old French had a ‘Germanic-like’ V2 syntax. Sitaridou (2012: 592) reaches a different conclusion, namely that Old French was a V2 language but that Old Occitan was not. The core argumentation against the V2 account of Old French and Old Occitan comes from the attestation of V1 and V3* orders, which are viewed as incompatible with a V2 grammar. Note for example the observation by Rinke and Elsig (2010: 2566) that ‘[a]s regards verb-initial and verb-third orders, they are clearly not a freely available option of verb-second grammars’. In our discussion, we will argue that both the Old French and Old Occitan data examined do, in fact, lead to a V2 analysis, but that subtle distinctions in the types of V1 and V3* order licensed are indicative of an important distinction between the grammars. 1.2.2 Questions in the V2 typology We do not attempt a full review of the vast literature on the V2 property here, but note one particularly important point for the discussion that follows. This concerns the nature of the left periphery. Seminal work on the V2 property for the most part assumed a single CP-projection (Den Besten 1983, 1989; Holmberg and Platzack 1995; Vikner 1995), which straightforwardly derived the occurrence of the verb in linear-second position (7): (7)

[CP Denne bog [C⁰ har] [IP Peter læst]] this book has Peter read.PTCP “Peter has read this book” (Modern Danish, Vikner 1995: 39)

The advent of cartographic approaches to the left periphery, which assume a far more articulated layering of functional projections (see for example Rizzi 1997; Poletto 2000; Benincà and Poletto 2004 and Ledgeway 2010) pose a number of intriguing questions, however. Are the full range of left-peripheral projections assumed in much cartographic work available in all V2 languages? If so, how do we account for the fact that orders where the verb is in third, fourth or even fifth position are licit in some V2 languages (see for example Benincà 2013 on Old ItaloRomance) but not in others (see Holmberg 2015:§4-5)? One potential answer to this question, which we pursue in some detail here with reference to Old Gallo-Romance, is found in Poletto (2002). Poletto proposes that there are at least two classes of V2 languages. In

58

Chapter Three

Force-V2 languages, finite verb movement and ‘EPP-effects’ target Force⁰ and Spec-ForceP respectively (8a), in addition to Fin⁰ and Spec-FinP. This yields a descriptively strict V2 system of the type found in most Modern Germanic varieties (Vikner 1995; Haider 2010: 1; Holmberg 2015:§5), where there is little or no functional structure c-commanding the finite verb and merged constituent satisfying the V2 constraint. By contrast, in Fin-V2 languages, the finite verb and EPP-effects are located on the lowest of the C-related functional projections, predicting the widespread attestation of V3* orders due to the rich functional structure ccommanding the moved verb and merged V2 constituent (8b). (8)

a. b.

[FrameP [ForceP V2 Constituent [Force⁰ VFin …]]] [FrameP…[ForceP… [TopP… [FocP… [FinP V2 Constituent [Fin⁰ VFin][TP…]]]]]]

Following the study of Rouveret (2004), which suggests that later Old French belongs in the ‘Force-V2’ category, we apply this insight to a comparative approach. Based on a small scale quantitative study of two texts, we propose that the Fin/Force dichotomy can in fact shed new light on the issue of how Old Occitan word order relates to that of its sister language, Old French. This in turn yields new perspectives on just how similar these two languages were in the medieval period and how the core properties of the grammar may have evolved diachronically. 1.2.3 Textual sample In what follows, we use as our starting point an analysis of a 1000clause sample of two texts. In the case of Occitan, we examine La Vie de Sainte Douceline, a piece of religious prose which was composed during the 13th century and has previously been employed by Vance, Donaldson and Steiner (2009) in their discussion of Old Occitan word order. We also make use of the well-studied early 13th century text La Quête du Saint Graal as representative of one variety of later Old French, which like Douceline is a religious narrative prose text.

2. Evidence for a V2 grammar When considered side by side, both languages present compelling evidence that movement of the finite verb targets the C-domain, rather than a functional projection in the T-layer, as is the case in unmarked matrix declaratives in Modern French and Occitan varieties (see Schifano

Syntactic Variation in Two Sister Languages

59

2015:Ch3 on V-to-T⁰ movement in French). Our investigation therefore confirms the long-held intuition that earlier French was a form of V2 system and more recent investigations into Occitan which have put forward similar proposals (cf. in particular Vance, Donaldson and Steiner 2009). The first piece of evidence for V-movement into the C-domain comes from data collected on the position of the finite verb in the linear ordering. Crucially, in V2 languages, the two properties of V-to-C movement and the presence of a phrasal movement-triggering feature on a C-head ‘conspire’ to yield a grammar where second position is unmarked for the finite verb._ As noted above, there is considerable variation between varieties where there is clear evidence for both the features responsible for V2 as to the frequency with which the verb occurs in second position and this is the case when the data in Tables 1 and 2 are examined from our French and Occitan texts:

Chapter Three

Total Percentage

Transitive/Unergative Unaccusative Reflexive Impersonal Athematic Copula Raising

V2 171 24 13 2 54 63 1 328 52.73%

V1 20 8 4 0 9 6 0 47 7.56%

185 29.74%

V3 79 14 19 0 49 24 0 50 8.04%

V4 17 4 4 1 19 5 0 8 1.29%

V5 5 1 0 0 2 0 0

Table 1: Verb Placement in Old Occitan Matrix Clauses (Excluding Coordination)

60

4 0.64%

V6 2 0 2 0 0 0 0 0 0.00%

V7 0 0 0 0 0 0 0 Total 622

Total Percentage

Transitive/Unergative Unaccusative Reflexive Impersonal Athematic Copula Raising

V2 250 53 17 1 85 65 4 475 75.16%

V1 0 0 0 0 0 0 0

0 0.00%

155 24.53%

V3 76 28 17 1 22 11 0 2 0.32%

V4 0 0 0 0 2 0 0 0 0.00%

V5 0 0 0 0 0 0 0

Table 2: Verb Position in Old French Matrix Clauses (Excluding Coordination)

Syntactic Variation in Two Sister Languages

0 0.00%

V6 0 0 0 0 0 0 0 0 0.00%

V7 0 0 0 0 0 0 0 Total 632

61

62

Chapter Three

At this stage we simply note that the data from verb placement are suggestive of a V2 grammar, but by no means conclusive. It is potentially significant to note that second position is the dominant order in both texts, but we should also note the revealing differences between both sets of data. Whereas in the French text no V1 in matrix declaratives is attested, this accounts for a robust 7.56% of the Occitan matrix sample. A similar pattern obtains with V4* orders, which are near-absent in French but account for 9.97% of Occitan matrix clauses. An analysis of the constituents appearing before the finite verb in matrix V2 clauses is more revealing. Here there is an opportunity to test a notable difference between SVO and V2 languages, as in the former class of languages the preverbal field is a specialised subject position. On the contrary, in V2 systems, it is well established in the literature that a variety of constituents from a range of grammatical classes and an array of discourse-pragmatic values can occur before the finite verb (Lightfoot 1995: 40; Frey 2004: 3; Westergaard 2008: 1843). This is indeed the case in both French and Occitan, as expected based on previous studies. (9)

a.

b.

c.

Aquesta obedientia de caritat this obedience of charity tenc illi tant cant le hold.3SG.PST her such that her paires visquet father live.3SG.PST “She undertook this exercise in charity such that her father lived” (Old Occitan, Douceline 47) Acostumat avia li de accustom.PTCP have.3SG.PST the sancta de pagar a Dieu saint to pay.INF to God las horas the hours “The saint had become used to reciting her hours to God” (Old Occitan, Douceline 159) e sens aquesta res non and without this thing NEG pot plazer a Dieu can.3SG.PST please.INF to God “and without this thing she could not please God” (Old Occitan, Douceline 69)

Syntactic Variation in Two Sister Languages

d.

e.

f.

63

Et de cel lac issoient and of that lake come-out.3PL nuef flum dont li uit nine rivers of-which the eight estoient d’une grandor et d’une be.3PL.PST of-a size and of-a parfondece depth “And nine rivers emerge from that lake, of which eight are large and deep” (Old French, Quête 135) et a chascun rendra ce and to each give.3SG.FUT that qu’ il avra deservi which he have.3SG.FUT deserve.PTCP “And he’ll give to everyone what he deserves…” (Old French, Quête 124) Et por ce vos pre je and for that you=ask.1SG I “And because of this, I ask you…” (Old French, Quête 152)

A sceptic may be tempted to dismiss the occurrence of such clauses as marginal. Such a stance, however, would not fit with the data. XPNon-SubjectVFin-(S) clauses outnumber SVO clauses both in the case of French (53.68% vs. 46.32//220 vs. 255) and to an even more substantial degree in the case of Occitan (76.22% vs. 23.78%//78 vs. 250). O-V-(S) also accounts for 35.98% of the Occitan matrix clauses examined (118/328) and 17.47% of the French matrix clauses examined (83/475). For now, the key point to note is that that these patterns are seemingly unexpected under an SVO analysis, but we will return to consider the variation between French and Occitan in this regard below. Our final piece of evidence from matrix clauses in favour of a V2 analysis comes from ‘inversion’ structures with postverbal subjects. Unsurprisingly, La Quête shows Germanic-inversion structures which unambiguously demonstrate V-to-C movement (10), as reported by other scholars (cf. references in §1.2.1). A novel finding, however, is that Occitan also shows such orders where a postverbal subject occurs between a finite auxiliary and an element unambiguously demarcating the leftmost boundary of the v-VP complex (11):

64

(10)

Chapter Three

a.

b.

(11)

a.

b.

Et toutes ces choses t’ ai and all these things you=have.1SG je dites I say.PTCP “and I have told you all these things…” (Old French, Quête 127) Son cors ne poï je his body NEG can.1SG I veoir see.INF “I cannot see his body” (Old French, Quête 122) Acostumat avia li Sancta accustom.PTCP have.3SG.PST the saint per amor del Seinnhor, de for love of-the Lord to far aquellas obras de caritat am do.INF those works of charity at gran compassion great compassion “The Saint had got used to doing those works of charity with great kindness, through her love of the Lord” (Old Occitan, Douceline 66) La qual cauza plus fizelmens the which thing more faithfully a far e plus veraia to do.INF and more truly volc illi aver per lo want.3SG.PST she have.INF for the dechat e-l conseill dell saint words and-theadvice of-the holy paire father “In order to this more truthfully, she wanted to have the words and guidance of the Holy Father” (Old Occitan, Douceline 61)

To summarise, we have noted clear evidence for a V2 syntax in both texts, based on the nature of the preverbal field and the inversion structures licensed. We also saw above that both texts display a strong tendency for the finite verb to occur in second position, though noted variation between the texts concerning the acceptability of V1 and V4* orders.

Syntactic Variation in Two Sister Languages

65

3. The structure of the left periphery 3.1 Discourse-pragmatic characteristics of preverbal constituents In our discussion of the diverse grammatical categories of constituents that can occupy the verbal prefield in both Old French and Old Occitan we noted an intriguing difference between the textual samples. It was shown that both XPNon-Subject-VFin-(S) orders in general and O-VFin-(S) orders specifically are more numerous in Occitan than in French. Here we put forward the tentative hypothesis that this may be due to differences in the syntax-Information Structure mapping in the left periphery of the languages in question. A qualitative analysis of both texts reveal that both languages license thematic elements in the preverbal field (12). These constituents occupy a high position on Lambrecht’s (1994: 165) Topic Acceptability Scale, are frequently active in the discourse (Prince 1981: 243) and and often correspond to ‘what the sentence is about’ (see Reinhart 1981: 58-60, Gundel 1988: 210 and Frascarelli and Hinterhölzl 2007:§1 on ‘aboutness’). Both texts also show clear evidence that they can license Frame-Setters in the left-periphery (13), which typically occur alongside other left-peripheral elements. These typically anchor the speech-act temporally, spatially or aspectually and scope over the entire utterance (see Chafe 1976: 50 et seq.) (12)

a.

b.

(13)

a.

Aquesta obedientia de caritat this obedience of charity tenc illi tant cant le keep.3SG.PST she such that the paires visquet father live.3SG.PST “She kept this obedience to charity such that her father lived” (Old Occitan, Douceline 47) Il oste ses armes he remove.3SG his arms “He removes his weapons” (Old French, Quête 122) e, [cant o ac auzit], and when it=have.3SG.PST hear.PTCP [li vera amairis de tota puritat] the true lover of all purity

66

Chapter Three

b.

menet… feel.3SG.PST “and when she learnt of this, the lover of absolute purity felt…” (Old Occitan, Douceline 95-96) Et [quant il est apareilliez, and when he be.3SG appear.PTCP il prent ses armes he take.3SG his weapons et monte and ride.3SG “When he appeared, he took his weapons and rode…” (Old French, Quête 129)

There is a clear point of variation between the Occitan and French texts concerning the realisation of preverbal Information Focus. Following Cruschina (2006, 2008, 2012) we understand such examples as involving foci which are non-contrastive yet informationally new. The Occitan text shows ample evidence for a preverbal projection hosting Information Focus (14), thus patterning with a number of Old Italo-Romance varieties where similar patterns are reported (Benincà 2004: 267; Ledgeway 2007: 130-131; Poletto 2014: 10):1 (14)

a.

b.

e voluntiers fazia lur and voluntarily do.3SG.PST their mandament order “and she did what they commanded voluntarily” (Old Occitan, Douceline 44) Mortification de carn comenset mortification of flesh begin.3SG.PST a penre tantost to take.INF early “She began to flagellate herself early [in the morning]” (Old Occitan, Douceline 48)

The same cannot be said for La Quête. There is no clear evidence of a preverbal position dedicated to Information Focus, most tellingly indicated by the fact that all fifty initial direct objects within the sample represent 1

Left-peripheral Information Focus is also licensed in Modern Sicilian (Cruschina 2006, 2012), Sardinian (Mensching and Remberger 2010) and Romanian (Ledgeway 2012: 158).

Syntactic Variation in Two Sister Languages

67

discourse-active information. The observation that the preverbal position increasingly shows a tendency to host discourse ‘old’ constituents throughout the Old French period is not new (Marchello-Nizia 1995: 99f; Vance 1997: 57; Rinke and Meisel 2009: 112; Labelle and Hirschbühler forthcoming: 20; Zimmerman 2014: 141). However, the fact that there is a clear point of contrast between La Quête and an Occitan text of a similar period is a novel observation and a significant one given the importance attached to the realisation of Focus in medieval texts across the relevant literature (see Benincà 2004, 2006, 2013 and Poletto 2014 amongst many others).

3.2 Verb third and the Fin/Force distinction Recall from our discussion above that the type of V3* order licensed within the texts is significant in revealing where the finite verb lands in the left periphery of the languages in question. As we shall see, there is a significant difference between our texts in this regard. Looking first at Occitan, we observe that the 29.74% of matrix clauses which show a V3 order fall into two major categories. Firstly, we observe orders where a preverbal Topic or Focus co-occurs with a Frame-Setter (15). This may in fact be a V3 pattern attested across the Medieval Romance languages in all periods (see Wolfe 2016 for the relevant data). (15)

a.

b.

[Cascun jorn], [aquist verge] annet each day this virgin go.3SG.PST de ben en miels of well on better “Each day the virgin kept getting better and better” (Old Occitan, Douceline 44) [E cant avia una pessa and when had.3SG.PST a little annat cantant], [illi] go.PTCP sing.PROG she se restancava apres SELF=rest.3SG.PST after “And when she had sung a little, she rested afterwards” (Old Occitan, Douceline 144)

Secondly, the hallmark orders of a Fin-V2 system are licensed in the text. Observe in (16) matrix clauses where a discourse-old Topic and new Information Focus co-occur:

68

(16)

Chapter Three

a.

b.

[Li segona lumniera], [non mens the second light NEG less luzens per sanctitat de vida], bright for sanctity of life fon ma dona sancta be.3SG.PST my lady saint Doucelina de Dinha Douceline of Digne “The second light, no less bright due to the holiness of her life, was Saint Douceline of Digne” (Old Occitan, Douceline 43) [Illi], [per amor del Senhor], she for love of-the Lord lur lavava los pes them=wash.3SG.PST the feet “Through her love of the Lord, she washed their feet” (Old Occitan, Douceline 45)

Further evidence for the Fin-V2 hypothesis comes from V4* clauses such as the following where three or more constituents occur before the finite verb. Crucially, these suggest that a rich range of left-peripheral projections can be targeted by maximal constituents (referred to as the ‘multiple accessibility of CP’ by Benincà 2004: 275 with reference to Medieval Italo-Romance): (17)

a.

b. ͒

E [per aisso], [illi] [adoncs], and for this she therefore [am gran confusion], at great confusion comandet a totas command.3SG.PST to all “Because of this, amongst great confusion, she commanded everyone to…” (Old Occitan, Douceline 130) E [adoncs] [illi], [ab amars and thus she with bitter critz], dizia a la Verge cries say.3SG.PST to the Virgin “and she cried in a bitter tone to the Virgin…” (Old Occitan, Douceline 136)

Syntactic Variation in Two Sister Languages

c.

69

[Per aisso], [aquist maire sancta, ma for this this mother saint my donna Doucelina], [tot aquest mont] lady Douceline all this world mesprezet despise.3SG.PST “and because of this, the mother saint Douceline hated the whole world” (Old Occitan, Douceline 76)

The overall generalisation for Occitan is that left-peripheral FrameSetters occupying the upper reaches of the left periphery can co-occur with Topics and Foci in the lower ‘fields’ of functional projections. A Topic and Focus can also co-occur and in this regard the Old Occitan text patterns closely with Old Italo-Romance texts of the 13th century (see Benincà 2004 and references above). Turning to the French text, we also find that constituents occupying the upper fields of the left periphery can co-occur with an additional leftperipheral constituent, typically a Topic. This most standardly occurs with a frame-setting (18a) or speaker-oriented adverbial expression (18b) and less commonly with a Hanging Topic, which is resumed by a pronominal form (18c): (18)

a.

b.

Et [quant il est apareilliez], and when he be.3SG appear.PTCP [il] prent ses armes et he take.3SG his weapons and monte ride.3SG.PST “When he appeared, he took his weapons and rode…” (Old French, Quête 129) Et [neporec] [Nostre Sires] and nevertheless our Lord avoit mis tant de has.3SG.PST put.PTCP such of bien en toi… good in you “And nevertheless our Lord has put so much good in you…” (Old French, Quête 126)

70

Chapter Three

c.

[Li chevalier qui sont en the knights that be.3PL in pechié mortel], ce sont li sin mortal they be.3PL the terrien… earthly… “the knights who are mortal sinners, they are the earthly ones” (Old French, Quête 143)

All the initial elements in these V3 constructions have in common that they would standardly be analysed as base-generated in the upper projections of the C-layer._ This would then be compatible with the ForceV2 analysis of later Old French proposed by Rouveret (2004), where only the functional projections above Force are able to host additional constituents to the one satisfying V2: (19)

[FrameP quant il est apareilliez [ForceP il [Force⁰ prent …]]]

If this hypothesis is correct, it predicts that in the case of La Quête V4* orders should be near-entirely absent, as should Topic-Focus orders. This is indeed correct. Table 2 shows that V4* orders account for a mere 0.32% of the matrix corpus and whilst Topic-Focus orders are widely attested in Early Old French verse (see Labelle 2007: 303; Donaldson 2012: 1038 and Mathieu 2012: 339-341), they are not present in La Quête. Our proposal is therefore that the grammatical system attested in La Quête is an attestation of a Force-V2 system. Note that although this proposal makes use of the conceptual tools provided by recent cartographic developments concerning the syntax of the left periphery, the core intuition behind it, that there is a an additional ‘layer’ for constituents which precede the immediately preverbal position, has a very long pedigree. Note for example, Skårup’s (1975: 179) proposal for an ‘extraposition’ layer before the immediately preverbal field. Although many subsequent proposals for the syntax of (later) Old French have differed markedly in specific details, it is perhaps no coincidence that many only posit two positions able to host constituents before the finite verb (Ferraresi and Goldbach 2002; Labelle and Hirschbühler 2005; Labelle 2007: 302; Simonenko and Hirschbühler 2012; Salvesen 2013). In comparing our two texts, our novel hypothesis is that the Old Occitan system instantiated in Douceline differs from La Quête differs not in its V2 status tout court, but in the fact the former is a Fin-V2 system and the latter a Force-V2 system. We will now see how this hypothesis permits

Syntactic Variation in Two Sister Languages

71

us to account for further morphosyntactic differences between the texts in a novel fashion.

3.3 Verb first In the previous section we saw that a significant difference in the type of verb-third orders licensed in our two texts can be accounted for under the hypothesis that the Old Occitan V2 grammar under consideration has access to the full range of left-peripheral projections, whilst this is not the case for the later Old French V2 grammar we consider here. The data in Table 1 and 2 are suggestive of a further difference between the texts concerning the licensing of V1 orders in matrix declaratives. While 7.56% of the Occitan matrix corpus shows a V1 order (47/622), declarative V1 in matrix clauses is entirely absent in the French text. Note that this pattern is by no means atypical of the wider textual record of both languages. Scholars have long noted that clauses with the verb in initial position become extremely rare in 13th century French prose (Skårup 1975: 291; Marchello-Nizia 1980: 331; Rouveret 2004: 193-195; Labelle and Hirschbühler 2005: 66), whilst the relatively widespread attestation of V1 clauses as a marked word order in many Old Occitan texts of the same period is equally well established (Jensen 1994: 359-360; Kunert 2003: 200; Sitaridou 2012: 570). Here we adopt a version of the hypothesis put forward by Benincà (2004: 290, 2006: 69), Ledgeway (2008: 442), Salvi (2012: 106-107) and Poletto (2014: 21) that V1 matrix clauses in Medieval Romance involve a phonologically null element, which we take to be a sub-type of the null pronoun pro licensed in Null Subject Languages._ Given that these clauses in Old Occitan are licensed either in contexts of topic continuity (20a) or rhematicity (20b), I take this proTop in line with Poletto (2014: 21) to act as a null variant of a Shift Topic which thus hosts the D-feature and φfeatures borne by pro in canonical Null Subject Languages (Roberts 2010), with the sole addition of an unvalued Topic feature. (20)

a.

Corregron tantost après per run.3PL.PST soon after to seguir las follow. INF=them “They ran soon after to follow them” (Old Occitan, Douceline 54)

72

Chapter Three

b.

Era oracion en totas cauzas be.3SG.PST prayer in all things sos refugz her refuge “Prayer was her refuge in all things” (Old Occitan, Douceline 106)

Although the main focus of the paper is not theoretical, I put forward the proposal which has been developed in more detail elsewhere (Wolfe 2016) that the licensing of proTop is contingent upon this pronoun raising from the clausal core to the Topic layer in the left-periphery, in order that the Top⁰ head can value its unvalued topic feature. This, I suggest, gives us a novel explanation as to why V1 clauses are widespread in Occitan texts of the type discussed here, but not in the majority of later Old French texts. As a Fin-V2 language, the Topic-layer makes up one set of the functional projections which constitute the verbal prefield. As such, a null topic whose eventual ‘destination’ within the clausal hierarchy is the Topic-layer can satisfy the V2 related EF as it moves through Spec-FinP and act as the sole constituent in the verbal prefield in a V2 clause. The Force-V2 system present in later Old French texts, however, does not have this option as the Topic-layer is ‘lower’ in the structure than the projections which make up the preverbal field (Frame and Force). Preverbal null topics are therefore ruled out. Put more simply, the licensing of V1 clauses is contingent upon a null topic reaching a position in the derivation which is structurally higher than the final landing site of the finite verb. This is possible in FinV2 systems, but not in Force-V2 systems. Our proposal is therefore that the Fin/Force distinction is not only crucial to understanding the patterns of V3* orders attested (as argued by Poletto 2002 for Rhaeto-Romance) but also the types of V1 licensed in the system.

3.4 The status of SI The final topic we consider within the realm of matrix clause syntax is the status of the particle SI, which is present in our French text but not in the Occitan text examined here. This element, derived from Latin SIC ‘thus’ is abundant in the textual records of a range of Medieval Romance languages and has attracted a vast literature. Due to space constraints, we adopt the hypothesis without discussion here that SI in Medieval Romance acts as a form of last-resort mechanism to satisfy the Edge Feature

Syntactic Variation in Two Sister Languages

73

associated with the V2 property, when no suitable XP constituent can.2 As such, clauses with SI such as that in (21) will feature the V-to-C movement associated with a standard V2 clause, but no phrasal movement: (21)

Si SI

osta sa take-off.3SG.PST his

chemise de shirt from

son his

dos back “He took off his shirt from his back” (Old French, Quête 121) There are two significant points to make about the structural distribution of SI within La Quête. Firstly, all cases of SI show a strict adjacency between the particle and the finite verb (Adams 1987b; Lemieux and Dupuis 1995: 96; Salvesen 2013: 142). Secondly, as noted by Marchello-Nizia (1985: 48), Van Reenen and Schøsler (1993: 617, 2000: 86) and Vance (1997: 53) amongst others, the co-occurrence of SI with overt subjects is rare, a finding confirmed by our corpus where only 2/64 (3.13%) SI-initial clauses contain an overt postverbal subject and only one clause contains an overt preverbal subject. Considering that subjects frequently act as the ‘default topic’ of a V2 clause, this alone is strong evidence that SI can be viewed as an alternative strategy to satisfy the Edge Feature on the head responsible for V2, in the absence of another topical or focal XP. The Fin/Force-V2 typology argued for in this chapter, however, has the potential to shed new light on the distribution of this particle. If, as we have seen above, Force acts as the eventual locus of the V2 property in the later Old French left periphery, a topical XP standardly satisfying V2 will be merged in Spec-ForceP. If SI provides an alternative strategy to satisfy V2, we make the strong prediction that it will have the same structural distribution as other constituents occupying this position. In particular, we predict that the exact same class of V3 configurations licensed when a topical XP is in Spec-ForceP will also occur with SI. Within La Quête this is precisely what we find. The matrix SI clauses within the corpus occur with an initial Frame-Setter (78/145), Hanging Topic (3/145) or have SI as the sole initial constituent (64/145), as exemplified in (22-24):

2

This last-resort analysis is in the spirit of Ledgeway (2008), though different in that merger of SI is not viewed as an alternative to V-to-C movement but rather XPmovement.

74

(22)

(23)

(24)

Chapter Three

Quand il vint en la valee, when he come.3SG.PST in the valley si comença a penser SI begin.3SG.PST to think.INF mout durement very hard “When he came to the valley, he began to think very hard” (Old French, Quête 145) Ce que messires Gauvains that what my-lord Gauvain vit en son dormant, see.3SG.PST in his sleep si li fu avis qu’il SI him=be.3SG.PST show.PTCP that-he ert en une pré plein be.3SG.PST in a meadow full d’erbe vert of-grass green “What my lord Gauvain saw in his sleep, he was shown to be in a meadow, full of green grass…” (Old French, Quête 149) Si lace son hiaume SI fasten.3SG his helmet “He fastened his helmet” (Old French, Quête 132)

Recall from the analysis of verb-third above that initial Frame-Setters and Hanging Topics are two of the three environments which license V3 in matrix clauses not featuring SI. Consistent with the Force-V2 hypothesis, we therefore propose that the later Old French variety instantiated in La Quête features a variant of SI which occupies Spec-ForceP: (25)

[FrameP (Frame-Setter/HT) [ForceP SI [Force⁰ VFin …]]]

We finish with a brief speculation concerning SI in Old Occitan. If Occitan differs from the later Old French variety instantiated in La Quête in having a structurally lower locus of V2, on Fin, we predict SI to have a distinct distribution from its counterpart found in La Quête. Specifically, we expect SI to be readily preceded by multiple left-peripheral constituents in the Frame-Topic-Focus layer, which is not accessible in La Quête. We cannot test this hypothesis with data from Douceline, as this variety does not feature SI. Crucially, Old Occitan varieties attested in the Vidas,

Syntactic Variation in Two Sister Languages

75

troubadour bibliographies from the 13th and 13th century, do feature SI and show precisely the distribution that the Fin-V2 analysis would predict. Observe the following data which show both SI-initial clauses and clauses where SI is preceded by other left-peripheral constituents: (26)

a.

b.

c.

Guillems de la Tor si Guillem de la Tor si fon joglars… be.3SG.PST jongleur “Guillem de la Tor was a jongleur…” (Old Occitan, Vidas, XXXII) E en aqesta [a]legressa, lo and in this happiness the marques de Monferrat si marquis of Monferrat SI se croset REFL= cross.3SG.PST “In the midst of this happiness, the Marquis of Moferrat crossed himself” (Old Occitan, Vidas, 167,33) Bertrans, quan auzi so Bertran when hear.3SG.PST that qu’en Richartz avia jurat, which Richard have.3SG.PST swear.PTCP e sabia qu’el era and know.3SG.PST that=he be.3SG.PST abandonatz de totz aquestz que abandon.PTCP of all these that vos avetz ausit, si you have.2PL hear.PTCP SI l det lo castel e him=give.3SG.PST the castle and si venc a son comandamen SI come.3SG.PST to his service “When Bertran heard what Richard had sworn, knowing that he had been betrayed in all these things about which you have heard, he gave him his castle and rendered himself at his service” (Old Occitan, Vidas, 80, 21)

76

Chapter Three

The full range of relevant facts are found in Donaldson (2015, 2016) who provides a full discourse-pragmatic analysis of SI in Old Occitan texts. For our purposes, however, we make the novel claim that the distribution of SI in 13th century Occitan texts is not the same as that found in La Quête and that this is linked to the fact that the grammatical systems under consideration have a different locus of V2.

4. Summary and conclusions 4.1 The Old Occitan and Old French C-domain in a comparative perspective This short study has shown how a systematic, side-by-side comparison of a 13th century Occitan text and a 13th century French text points to a number of revealing similarities and differences between the two languages. We have seen that both texts provide evidence that strongly points towards a V2 analysis. Both texts show a strong preference for a linear-V2 order, have a preverbal field which is evidently not a specialised subject position and, most compellingly of all, show inversion structures which cannot be accounted for by assuming V-to-T⁰ movement. We can therefore conclude that having finite verb-movement and EPP-effects in the Cdomain is a point of commonality between these two medieval GalloRomance varieties. Beyond this point of continuity, however, there are points of contrast. Whilst both texts show evidence of a verbal prefield that can host topical and contrastively focussed constituents, only Occitan Douceline provides unambiguous evidence of a preverbal position for new Information Focus. The texts differ also in the types of V3 orders they license, with Douceline showing clear attestations of V>4 orders, in contrast to La Quête and evidence that projections with the Frame-Topic-Focus field can be simultaneously activated. In our later Old French text, La Quête, by contrast, V3 is far more restricted to occurring with an initial Hanging Topic or Frame-Setter. In a similar vein, the V2 expletive particle SI has a different distribution in both texts, readily occurring with multiple constituents in the Frame-Topic-Focus layer in the Occitan text, yet only co-occurring with Hanging Topics and Frame-Setters in the French text. Finally, we noted an important point of contrast regarding the null argument properties in the texts. Whilst the Occitan text, Douceline, shows a robust attestation of V1 and is thus analysed as licensing initial Null Topics, the later Old French text examined shows no V1 matrix

Syntactic Variation in Two Sister Languages

77

declaratives and thus seems restricted to licensing null arguments exclusively in postverbal position. Our analysis here is that there is a common thread that accounts for these points of contrast, namely that the evidence examined points to the texts as being instantiations of different types of the V2 system. In the Occitan text, V-movement and EPP-effects target Fin⁰ and Spec-FinP respectively. Consequently, the Occitan variety under consideration has a richly articulated verbal prefield made up of a range of functional projections. By contrast, in the French variety we consider here, Vmovement and EPP-effects are additionally associated with Force⁰ and Spec-ForceP. As such the verbal prefield is structurally more impoverished than in Occitan, yielding a descriptively richer V2 system. We can schematise these findings, drawing on earlier insights of Rizzi (1997 et seq) and Benincà and Poletto (2004) as in (27). Moving rightwards through the hierarchy of functional projections, we see that both systems feature a Frame field which hosts Frame-Setting clauses and adverbials along with base-generated Hanging Topics. The Force field below in French only hosts the finite verb, the constituent merged to satisfy the V2-related EF and the expletive particle SI. Below this we have the Topic-field, which in Occitan alone can host a topical null pronoun yielding V1 clauses, followed by the Focus-field which only in Occitan can host Information Focus. Finally, we come to FinP, which acts as a bottleneck for merged constituents and is targeted by verb movement in both languages, but acts as the eventual locus of SI only in Occitan. (27)

[FrameP Frame-Setter/Hanging Topic [ForceP Merged Phrasal Constituent(French)/SI(French) [CForceº VFin(French)][TopP Topic/proTop [FocP Focus [FinP Merged Phrasal Constituent(French/Occitan)/SI(Occitan) [CFinº VFin(French/Occitan)]]]]]]

4.2 Implications of the research Overall the research presented here contributes to a wide variety of data-rich studies emerging on a number of Medieval Romance varieties. The growing picture, which the French and Occitan data discussed here support, is that the intuition in Benincà (1995, 2004, 2006, 2013) that the Medieval Romance languages shared the V2 property, stands up as correct under detailed scrutiny. However, there are rich patterns of microvariation evidenced across the textual records that suggest differences between sister languages should not be overlooked. It may be the case in the fullness of

78

Chapter Three

time that the growing body of theoretically-informed work on Medieval Romance is indicative of early levels of microvariation within the Romance family, which are equal to those found today across Romancespeaking Europe (see in particular Poletto 2000; Kayne 2000, 2005 and D’Alessandro, Ledgeway and Roberts 2010). Finally, we note that the findings presented here may be diachronically significant. Crucially, the available literature on the earliest 11th and 12th century French texts evidence an articulated left-peripheral structure entirely comparable with that sketched for Occitan in this chapter. Further research may therefore be needed to evaluate the hypothesis that the Gallo-Romance languages were at the turn of the last millennium more syntactically homogeneous than was later the case.

References Adams, M., 1987a,‘From Old French to the Theory of Pro-Drop’, Natural Language & Linguistic Theory 5 (1), 1–32. Adams, M., 1987b, ‘Old French, Null Subjects, and Verb Second Phenomena, Doctoral Dissertation, UCLA. Benincà, P., 1995, “Complement Clitics in Medieval Romance: The Tobler Mussafia Law”, in: I. Roberts, A. Battye (eds), Clause Structure and Language Change, Oxford, Oxford University Press, 325–344. Benincà, P., 2004, “The Left Periphery of Medieval Romance”, Studi Linguistici E Filologici Online, 2, 2, 243–297. Benincà, P., 2006, “A Detailed Map of the Left Periphery of Medieval Romance”, in: R. Zanuttini (ed.), Crosslinguistic Research in Syntax and Semantics: Negation, Tense and Clausal Architecture, Georgetown, Georgetown University Press, 53–86. Benincà, P., 2013, “Caratteristiche Del V2 Romanzo. Lingue Romanze Antiche, Ladino Dolomitico E Portoghese”, in: E. Bidese, F. Cognola (eds), Introduzione Alla Linguistica Del Mòcheno, Torino, Rosenberg and Sellier, 65–84. Benincà, P., C. Poletto, 2004, “Topic, Focus, and V2”, in: L. Rizzi (ed.), The Structure of CP and IP, Oxford, Oxford University Press, 52–75. Cardinaletti, A., I. Roberts, 2002, “Clause Structure and X-Second”, in: G. Cinque (ed.), Functional Structure in DP and IP, 1, The Cartograpy of Syntactic Structures, Oxford, Oxford University Press,123–166. Chafe, Wallace, 1976, ‘Givenness, Contrastiveness, Definiteness, Subjects, Topics and Point of View’, in Subject and Topic, edited by C. N. Li, New York, NY, Academic Press, 27–55. Cinque, Guglielmo, 1999, Adverbs and Functional Heads: A Cross-Linguistic Perspective, Oxford, Oxford University Press.

Syntactic Variation in Two Sister Languages

79

Cruschina, S., 2006, “Information Focus in Sicilian and the Left Periphery”, in: M. Frascarelli (ed.), Phases of Interpretation, Berlin, Mouton de Gruyter, 363–385. Cruschina, S., 2008, Discourse-Related Features and the Syntax of Peripheral Positions. A Comparative Study of Sicilian and Other Romance Languages, Doctoral Dissertation, Cambridge, University of Cambridge. Cruschina, S., 2012, Discourse-Related Features and Functional Projections, Oxford, New York, Oxford University Press. D’Alessandro, R., A. Ledgeway, I. G. Roberts (eds), 2010, Syntactic Variation: The Dialects of Italy, New York, Cambridge University Press. Den Besten, H., 1977, “On the Interaction of Root Transformations and Lexical Deletive Rules”, in: W. Abraham (ed.), On the Formal Syntax of the Westgermania, Amsterdam/Philadelphia, John Benjamins Publishing Company, 47–61. Den Besten, H., 1989, Studies in West Germanic Syntax, Amsterdam/Atlanta, Rodopi. Donaldson, B., 2012, “Initial Subordinate Clauses in Old French: Syntactic Variation and the Clausal Left Periphery”, Lingua, 122, 9, 1021–1046. Donaldson, B., 2015, “Discourse Functions of Subject Left Dislocation in Old Occitan”, Journal of Historical Pragmatics, 16, 2, 159–186. Donaldson, B., 2016, “Preverbal Subjects, Information Structure, and Object Clitic Position in Old Occitan”, Journal of Linguistics, 52, 1, 37–69. Ferraresi, G., M. Goldbach, 2002, “V2 Syntax and Topicalization in Old French”, Linguistiche Berichte, 189, 2–25. Frascarelli, M., R. Hinterhölzl, 2007, “Types of Topics in German and Italian”, in: S. Winkler, K. Schwabe (eds), On Information Structure, Meaning and Form, Amsterdam/Philadelphia, John Benjamins, 87–116. Frey, W., 2004, “The Grammar-Pragmatics Interface and the German Prefield”, Sprache Und Pragmatik, 52, 1–30. Gundel, J. K., 1988, “Universals of Topic-Comment Structure”, in: M. Hammond, E. Moravczik, J. Wirth (eds), Studies in Syntactic Typology, Amsterdam, John Benjamins, 209–239. Haider, H., 2010, The Syntax of German, Cambridge, Cambridge University Press. Hamlin, F. R., P. T. Ricketts, J. Hathaway, 1967, Introduction à L’étude de L’ancien Provençal, Geneva, Librairie Droz. Holmberg, A., 2000, “Scandinavian Stylistic Fronting: How Any Category Can Become an Expletive”, Linguistic Inquiry, 31, 3, 445–483. Holmberg, A., 2015, “Verb Second”, in: T. Kiss, A. Alexiadou (eds), Syntax— Theory and Analysis, Berlin, Walter de Gruyter, 242–283. Holmberg, A., C. Platzack, 1995, The Role of Inflection in Scandinavian Syntax, New York, Oxford University Press.. Hulk, A., A. van Kemenade, 1995, “Verb-Second, Pro-Drop, Functional Projections and Language Change”, in: I. Roberts, A. Battye (eds), Clause Structure and Language Change, Oxford, Oxford University Press, 227–256. Jensen, F., 1990, Old French and Comparative Gallo-Romance Syntax, Tübingen, Walter de Gruyter.

80

Chapter Three

Jensen, F., 1994, Syntaxe de L’ancien Occitan, Tübingen, Niemeyer. Kaiser, G. A., 2002, Verbstellung Und Verbstellungswandel in Den Romanischen Sprachen, Max Niemeyer Verlag. Kayne, R., 2000, Parameters and Universals, Oxford, Oxford University Press. Kayne, R., 2005, Movement and Silence, Oxford, Oxford University Press. Kunert, H., 2003, “L’ordine Degli Elementi Della Frase in Occitano Antico”, Romanische Forschungen, 115, 194–209. Labelle, M., 2007, “Clausal Architecture in Early Old French”, Lingua, 117, 1, 289–316. Labelle, M., P. Hirschbühler, 2005, “Changes in Clausal Structure and the Position of Clitics in Old French”, in: M.Batllori, M.-L. Hernanz, C. Picallo, F. Roca (eds), Grammaticalization and Parametric Variation, Oxford, Oxford University Press, 149–178. Labelle, M., P. Hirschbühler, Forthcoming, “Topic and Focus in Old French V1 and V2 Structures”, in: E. Mathieu, R. Truswell (eds), Micro-Change to Macro-Change in Diachronic Syntax, Oxford, Oxford University Press. Lambrecht, K., 1994, Information Structure and Sentence Form : Topic, Focus, and the Mental Representations of Discourse Referents, Cambridge: Cambridge University Press. Ledgeway, A., 2007, “Old Neapolitan Word Order: Some Initial Observations”, in: A. L. Lepschy, A. Tosi (eds), Histories and Dictionaries of the Languages of Italy, Ravenna, Longo, 121–149. Ledgeway, A., 2008, “Satisfying V2 in Early Romance: Merge vs. Move”, Journal of Linguistics, 44, 2, 437–470. Ledgeway, A., 2010, “Subject Licensing in CP: The Neapolitan Double-Subject Construction”, in: P. Benincà, N. Munaro (eds), Mapping the Left Periphery, Oxford, Oxford University Press, 257–296. Lemieux, M., F. Dupuis, 1995, “The Locus of Verb Movement in NonAsymmetric Verb-Second Languages: The Case of Middle French”, in: I. Roberts, A. Battye (eds), Clause Structure and Language Change, Oxford, Oxford University Press, 80–110. Lightfoot, D., 1995, “Why UG Needs a Learning Theory: Triggering Verb Movement”, in: A. Battye, I. Roberts (eds), Clause Structure and Language Change, Oxford, Oxford University Press, 31–52. Maling, J., 1990, “Inversion in Embedded Clauses in Modern Icelandic”, in: J. Maling, A. Zaenen (eds), The Syntax of Modern Icelandic, Syntax and Semantics 24, San Diego, Academic Press. Marchello-Nizia, C., 1980, La Langue Française Aux XIVe et XVe Siècles, Paris, Nathan. Marchello-Nizia, C., 1995, L’évolution Du Français: Ordre Des Mots, Démonstratifs, Accent Tonique, Paris, Armand Colin. Mathieu, É., 2006, “Stylistic Fronting in Old French”, Probus, 18, 219–266. Mathieu, É., 2009, “On the Germanic Properties of Old French”, in: Historical Syntax and Linguistic Theory, Oxford, Oxford University Press, 344–357. Mathieu, É., 2012, “The Left Periphery in Old French”, in: D. Arteaga (ed.), Research in Old French: The State of the Art, Dordrecht, Springer, 327–350.

Syntactic Variation in Two Sister Languages

81

Mensching, G., E.-M. Remberger, 2010, “Focus fronting and the left periphery in Sardinian”, in: R. D’Alessandro, A. Ledgeway, I. Roberts (eds), Syntactic Variation: The Dialects of Italy, Cambridge, Cambridge University Press, 261–276. Mussafia, A., 1888, “Enclisi O Proclisi Del Pronome Personale Atono Quale Oggeto”, Romania, 27, 145–146. Poletto, C., 2000, The Higher Functional Field: Evidence from Northern Italian Dialects. First Edition, Oxford, Oxford University Press. Poletto, C., 2002, “The Left-Periphery of V2-Rhaetoromance Dialects: A New View on V2 and V3”, in: S. Barbiers, L. Cornips, S. van der Kleij (eds), Syntactic Microvariation, Meertens Institute Electronic Publications in Linguistics 2, Amsterdam, Meertens Institute, 214–242. Poletto, C., 2014, Word Order in Old Italian, Oxford, Oxford University Press. Reenen, P. Th. van, L. Schøsler, 2000, “The Pragmatic Functions of the Old French Particles Ainz, Apres, Donc, Lors, Or, Puis and Si”, in: S. C Herring (ed.), Textual Parameters in Older Languages, Amsterdam, Philadelphia, John Benjamins, 59–105. Reenen, P. van, L. Schøsler, 1993, “SI “Thématique”, étude de SI En Ancien et Moyen Français, Discours Direct”, in: Actes Du XXe Congrès International de Linguistique et Philologie Romanes, Zürich, 617–628. Reinhart, T., 1981, “Pragmatics and Linguistics: An Analysis of Sentence Topics”, Philosophica, 27, 1, 53–93. Rinke, E., M. Elsig, 2010, “Quantitative Evidence and Diachronic Syntax”, Lingua, 120, 2557–2268. Rinke, E., J. Meisel, 2009, “Subject Inversion in Old French: Syntax and Information Structure”, in: G. A. Kaiser, E. M. Remberger (eds), Proceedings of the Workshop on Null Subjects, Expletives and Locatives in Romance, University of Konstanz, Fachbereich Sprachwissenschaft, 93–130. Rizzi, L., 1997, ‘The Fine Structure of the Left Periphery’, in: L. Haegeman (ed.), Elements of Grammar: Handbook of Generative Grammar, Dordrecht, Kluwer, 281–338. Roberts, I., 1993, Verbs and Diachronic Syntax: A Comparative History of English and French, Dordrecht, Kluwer Academic Publishers. Roberts, I., 2007, Diachronic Syntax, Oxford, Oxford University Press. Roberts, I., 2010, “A Deletion Analysis of Null Subjects”, in: T. Biberauer, A. Holmberg, I. Roberts, M. Sheehan (eds), Parametric Variation : Null Subjects in Minimalist Theory, by Cambridge, Cambridge University Press, 58–87. Rouveret, A., 2004, “Les Clitiques Pronominaux et La Périphérie Gauche En Ancien Français”, Bulletin de La Société de Linguistique de Paris, 99, 181– 237. Salvesen, C., 2011, “Stylistic Fronting and Remnant Movement in Old French”, in: J. Berns, H. Jacobs, T. Scheer (eds), Romance Languages and Linguistic Theory 2009: Selected Papers from ‘Going Romance’ Nice 2009, Amsterdam/Philadelphia, John Benjamins, 323–332.

82

Chapter Three

Salvesen, C., 2013, “Topics and the Left Periphery: A Comparison of Old French and Modern Germanic”, in: T. Lohndal (ed.), In Search of Universal Grammar: From Old Norse to Zoque, Amsterdam, John Benjamins, 131–172. Salvesen, C., K. Bech, 2014, “Postverbal Subjects in Old English and Old French”, Oslo Studies in Language, 6, 1, 201–228. Salvi, G., 2012, “On the Nature of the V2 System of Medieval Romance”, in: L. Brugè, A. Cardinaletti, G. Giusti, N. Munaro, C. Poletto (eds), The Cartography of Syntactic Structures. Vol. 7, Functional Heads, Oxford/New York, Oxford University Press, 103–111. Schifano, N., 2015, Verb Movement: A Pan-Romance Investigation, PhD Dissertation, University of Cambridge. Simonenko, A., P. Hirschbühler, 2012, “Placement de Clitiques Dans Les Propositions V1 et évolution de La Structure de La Proposition En Ancien Français”, in: M. Dufresne (ed.), Typologie, Ordre Des Mots et Groupe Verbal En Français Médiéval, Laval, Québec: Les Presses de L’Université Laval, 11–53. Sitaridou, I., 2005, “A Corpus-Based Study of Null Subjects in Old French and Old Occitan”, in: C. D. Putsch, J. Kabatek, W. Raible (eds), Corpora and Diachronic Linguistics, Tübingen, Narr, 359–374. Sitaridou, I., 2012, “A Comparative Study of Word Order in Old Romance”, Folia Linguistica 46, 2, 1–51. Skårup, P., 1975, Les Premières Zones de La Proposition En Ancien Français, København, Akademisk Forlag. Thurneysen, R., 1892, “Die Stellung Des Verbums in Altfranzösischen”, Zeitschrift Für Romanische Philologie, 16, 289–307. Tobler, A., 1875, “De L’ordre Des Mots Dans Chrétien de Troyes”, Vermischte Beiträge Zur Französischen Grammatik, 5, 395–414. Tobler, A., 1883, “Die Altvenzianische Ubersetzung Der Spruche Des Dionysius Cato”, AbhBerlin, 1–87. Vance, B., 1995, “On the Decline of Verb Movement to Comp in Old and Middle French”, in: A. Battye, I. Roberts (eds), Clause Structure and Language Change, Oxford, Oxford University Press, 173–199. Vance, B., 1997, Syntactic Change in Medieval French, Dordrecht, Kluwer. Vance, B., B. Donaldson, B. D. Steiner, 2009, “V2 Loss in Old French and Old Occitan: The Role of Fronted Clauses”, in: S. Colina, A. Olarrea, A. M. Carvalho (eds), Romance Linguistics 2009. Selected Papers from the 39th Linguistic Symposium on Romance Languages (LSRL), Tuscon, Arizona, Amsterdam, John Benjamins, 301–320. Vanelli, L., L. Renzi, P. Benincà, 1986, “Tipologia Dei Pronomi Soggetto Nelle Lingue Romanze Medievali”, Quaderni Patavini Di Linguistica, 5, 49–66. Vikner, S., 1995, Verb Movement and Expletive Subjects in the Germanic Languages, New York, Oxford University Press. Westergaard, M., 2008, “Acquisition and Change: On the Robustness of the Triggering Experience for Word Order Cues”, Lingua, 118, 12, 1841–1863.

Syntactic Variation in Two Sister Languages

83

Wolfe, S., 2016, “On the Left Periphery of V2 Languages”, Rivista Di Grammatica Generativa: Selected Papers from the 41st Incontro Di Grammatica Generativa, 38, 287–310.

CHAPTER FOUR SYNTACTIC ARCHAISMS PRESERVED IN A CONTEMPORARY ROMANCE VARIETY: INTERPOLATION AND SCRAMBLING IN OLD ROMANIAN AND ISTRO-ROMANIAN ADINA DRAGOMIRESCU, ALEXANDRU NICOLAE “Iorgu Iordan—Al. Rosetti” Institute of Linguistics / University of Bucharest [email protected]; [email protected]

1. Background, research questions and aims 1.1 Interpolation and scrambling: Old Romance, old Romanian, Istro-Romanian A well-known phenomenon attested in the older stages of the Romance languages is the discontiguity of the functional elements of the IP-domain, namely the existence of structures like: (A) [pron. clitic–XP– V] (= interpolation) (1) / (B) [(pron. clitic–)aux.–XP–lexical V] (= scrambling) (2); we will use the term “discontiguous structures” when we refer to both interpolation and scrambling. The phenomenon in (A) was discussed for the first time by the philologist W.H. Chénery (see Poole 2007) with reference to old Spanish; to Chénery we also owe the term interpolation. Subsequent research showed that similar phenomena (labelled as (IP-)scrambling, intercalation or dislocation of compound forms) exist in most of the other medieval Romance varieties (see Salvi 2001; Martins 2002; Fischer 2003; Poole 2007; Mensching 2012; Poletto 2014; Sitaridou 2012; Dragomirescu 2013 i.a.).

86

(1)

(2)

Chapter Four

E com todas as pertenças que aos and with all the belongings that to-the ditos casaes lhes mentioned properties them.DAT dereitamẽte pertẽçe (old Ptg.) rightly belongs “and with everything which according to the terms of the contract is a part of the mentioned properties” (apud Martins 2002) a. porque ella non avia las cartas because she not had.3SG the letters resçebidas (old Sp.) received “because she had not received the letters” (apud Mensching 2012) b. se l’avessi a mente if it-had.SUBJ.1SG in mind tenuto (old It.) kept “If I had kept it in mind” (apud Poletto 2014)

Interpolation and scrambling are also present in old Romanian (= ORom) (16th—18th c.) (Dragomirescu 2013, 2015); scrambling [(pron. clitic–)aux.–XP–lexical V] (3) is more frequently attested than interpolation [pron. clitic–XP–V] (4) (Nicolae 2015): (3)

(4)

că ne va pre that CL.ACC.1PL AUX.FUT.3SG DOM noi asculta (ORom) us listen.INF “that he will listen to us” (CCat.1560: 10r) b. cum au ei lăsat how AUX.PERF.3PL they leave.PPLE cu al lor cuvânt with AL.M.SG their word “how they said with their own words” (DÎ.1599:XXX) Deaca ne noi bolnăvim if CL.REFL.1PL we get.sick.1PL “if we get sick” (CC1.1567: 129v)

a.

Syntactic Archaisms Preserved in a Contemporary Romance Variety

87

These phenomena are also attested in Istro-Romanian (= IstroRom) (Zegrean 2012); the corpora we have examined reveal approximately the same distribution of discontiguous structures, with scrambling (5) being much more frequent than interpolation (6): (5)

(6)

che n-ŭai mire because not=AUX.PERF.2SG me.ACC scutat listen.PPLE “because you did not listen to me” (TC: 7) rei te tu AUX.FUT.2SG CL.REFL.2SG you.NOM marita dupa míre? marry.INF after me “Will you marry me?” (AK: 338)

(IstroRom)

(IstroRom)

1.2 Old Romanian, Istro-Romanian. Corpus examined ORom represents a period in the history of (Daco-)Romanian roughly stretching from the beginning of the 16th c. to the end of the 18th c. (Gheție 1975; Timotin 2016).1 ORom was employed in the region roughly corresponding to present-day Romania, the Republic of Moldova and a stretch of land from the southern Ukraine (northern Bukovina), which consisted in the Middle Ages of three (administratively separate) provinces: Transylvania, Wallachia (Rom. Țara Românească), and Moldova, which were subject to different linguistic and cultural influences: Hungarian, Old Church Slavonic, Medieval Latin, etc. (Stan 2013: ch. 1). The corpus on which our analysis is based consists of original texts (nontranslated texts written directly in Romanian) and translations from the entire area where ORom was used; scholars have stressed the importance of ‘original’ texts, considered to reflect the actual grammar of the language more faithfully than translations. IstroRom is an eastern Romance variety mainly spoken in Istria, a peninsula in Croatia (Coteanu 1957); the speakers are all bilingual in Croatian (Maiden 2016: 91). A large community of IstroRom émigré 1

The adjective “old” in the phrase “old Romanian” is somehow a misnomer when compared to its usage for other medieval Romance varieties, where “old” denotes a much older timespan; from a Romance perspective, “old Romanian” actually represents an early modern stage of this language (see also the discussion in Hill and Alboiu 2016:ch. 1). The label “old Romanian” is maintained in accordance with traditional scholarship.

88

Chapter Four

speakers is found in the New York area (Maiden 2016: 91). The most widely accepted hypothesis (see Pană Dindelegan 2013: 3–4 and references therein; Maiden 2016) is that IstroRom is a self-standing Romance language traditionally considered a “historical dialect” of Romanian. As shown by their phonological, morphological, syntactic and lexical commonalities, the Romanian “historical dialects” (DacoRomanian, Istro-Romanian, Aromanian, and Megleno-Romanian) once formed a unity (Proto-Romanian/Common Romanian/Primitive Romanian), a stage that can be examined only through comparative reconstruction). Shortly before the 10th c., a great dialectal separation occasioned by the Slavic settlement in the south of the Danube divided this unity into a northern Danubian area (subsequently Daco-Romanian) and a southern Danubian area (the other aforementioned varieties) (Vasiliu and IonescuRuxăndoiu 1986). The full separation of Aromanian is dated before the 11th c., while IstroRom did not undergo the split before the 13th c. (Maiden 2016: 91 and references therein). The corpus we have examined consists of IstroRom texts from the Istrian Peninsula, all of them collected after 1930; occasionally, material has been drawn from other studies on IstroRom.

1.3 Research questions and aims Against this background, the aims and research questions of our paper are twofold, descriptive and explanatory. (i) From a descriptive point of view, we endeavour to delineate the extension of the phenomena analysed on the basis of representative ORom and IstroRom corpora. From an explanatory perspective, we should provide answers to the following questions: (ii) What is the proper syntactic analysis of discontiguous structures in ORom and IstroRom? Poole (2007: 188) compares old Spanish and old Portuguese interpolation and concludes that: “«Interpolation» phenomena exist in other medieval Romance varieties, but seem to have different properties from interpolation in old Spanish, suggesting that different processes are at work” (italics AD&AN). Thus, in view of “Poole’s caveat”, the question should be reformulated as follows: (ii’) Should the ORom and IstroRom data be given a uniform syntactic analysis (whatever this analysis is)? Furthermore, should interpolation and scrambling (two phenomena usually kept distinct in the literature) be syntactically analysed as representing the same phenomenon?

Syntactic Archaisms Preserved in a Contemporary Romance Variety

89

The syntactic diagnostics we have applied point to this conclusion. This raises two further questions: (iii) What is the relation between ORom interpolation and IstroRom interpolation? Two options present themselves here: (a) we are dealing with independent convergent developments in both idioms, or (b) interpolation is an old Romance phenomenon, present in a nonattested stage of Romanian prior to the Daco-Romanian / IstroRom dialectal split and initially preserved in both idioms, which died out in the passage from old to modern Romanian,but was preserved as an archaic feature in Istro-Romanian. (iv) Is language contact relevant for interpolation? Due to consistent cultural and linguistic contact with (Old Church) Slavonic and Hungarian (and, to a more limited extent, Medieval Latin and Greek) (Stan 2013: ch. 1)—official administrative and/or liturgical languages in the medieval Romanian provinces from the north of the Danube—, scholars considered that interpolation and scrambling (usually discussed together) are a “nonRomanian syntactic feature” (e.g. Avram [1975] 2007: 94—Slavonic, Dragoș 1995: 37—Hungarian, i.a.). However, currently, with respect to ORom syntax, there is agreement that language contact actually consolidated the grammatical structure of the target language: “taking into consideration the extensive presence of these learnèd and «foreign» phenomena in Romanian original texts, we are led to conclude that they actually consolidate structures and constructions present in the grammar of Romanian, affecting their frequency, rather than resulting from a process of wholesale importation of foreign structures into the grammar” (Pană Dindelegan and Dragomirescu 2016: 636–637). The rich attestation of interpolation and scrambling in texts directly written in Romanian (Dragomirescu 2013) and their common medieval Romance ancestry (Dragomirescu 2015; Nicolae 2015) represent strong evidence against the “non-Romanian syntactic feature” hypothesis. In the same spirit, with reference to word order in general (not to interpolation/scrambling in particular), it has been claimed that Croatian had an important influence on IstroRom (Kovačec 1971: 174).

2. Interpolation in old Romanian and Istro-Romanian 2.1 Framework We first spell out the minimal theoretical ingredients necessary for the analysis of interpolation and scrambling in ORom and IstroRom. The following set of diagnostics is discussed on the basis of modern (Daco-)

90

Chapter Four

Romanian; it will be extended to ORom and IstroRom in the section devoted to their analysis. 2.1.1 The V-raising parameter In generative thinking, it is generally accepted that the clausal spine is divided into three domains: the vP-domain, the IP-domain and the CP-domain. We use the label “vP-domain” in a broad sense, namely to refer to the lowest domain of the clause, where thematic and predication relations are established. According to many authors, the clause-internal phasal domain—customarily labelled as the v*-phase (cf. Chomsky 2001) — contains more material than the minimal vP projection. Baltin (2012), Rouveret (2012) and Roberts (2013) assume that Voice rather than v is the clause-internal phasal head. For other authors the clause-internal phase is even larger, e.g. for Harwood (2015), the low Aspect projection (the projection responsible for progressive aspect in English) is also part of the v*-phase.2 V-raising can target either the IP-domain (e.g. many modern Romance varieties such as French and Italian) or the CP-domain (e.g. V2 in medieval Romance, German, Dutch and mainland Scandinavian), or it can be absent or very low (i.e. targeting heads on the vP-edge) (e.g. Latin, English) (7) (see Emonds 1978; Pollock 1989 for the English/Romance (French) contrast; den Besten 1983; Koster 1975 for Germanic V2; Salvi 2004; Ledgeway 2012 for Latin3). (7)

CP-domain Germanic V2

> IP-domain French, Italian

>

vP-domain Latin, English

Work in the wake of Cinque (1999) advocated a finer-grained structure of the IP-domain, with the IP being split at least into three distinct fields: IP → MoodP > TP > AspP. Recent research has shown that V-to-I movement is not a uniform phenomenon across Romance, and that 2

Building on Biberauer and Roberts (2010), Rouveret (2012) considers that the heads from the clause internal phasal domain are actually all varieties of little v, “each v is «non-distinct in formal features» from the other v’s” (p. 954). (All these heads can be even relabelled as vVoice, vAsp, etc.) 3 While root clauses display a mixture of pragmatically (and sometimes syntactically) driven verb fronting and SOV structures, embedded clauses (in which V-to-C is a priori potentially blocked) display a more rigid SOV grammar; this has been taken as a strong indication for the fact that the V-raising out of the vP-domain does not take place in Latin.

Syntactic Archaisms Preserved in a Contemporary Romance Variety

91

V-raising targets distinct positions in the complex split IP-domain (Ledgeway and Lombardi 2005; Ledgeway 2012: 140–150; Schifano 2015a, 2015b). Fleshing out this line of inquiry, Schifano identifies four distinct targets of V-to-I movement of synthetic verbs in Romance: (8)

MoodP > high (Mod.)Rom Fr.

TP > clause-medial N.R.It.

Asp low very low E.Ptg. Sp.

While Romance languages behave systematically with respect to the position targeted by the raising of the synthetic verb (Schifano 2015a, 2015b) (cf. (8)), there is a great deal of variation in Romance when it comes to analytic structures: languages patterning the same way in synthetic structures (e.g. French and Romanian, high verb movement languages in (8)) display a different behaviour in analytic structures. Witness the following contrast, which indicates that the lexical verb has undergone V-to-I movement and clusters together with the auxiliary verb in Romanian, but not in French, where occurrence to the right of modal (9) and manner (10) adverbs and of floating quantifiers (11) indicates that the French lexical verb does not undergo raising to the inflectional domain (Dobrovie-Sorin 1994: 8–9; Alboiu and Motapanyane 2000: 15; Ledgeway 2015: 13–17): (9)

a. b.

(10)

a. b.

(11)

a. b.

Il est probablement venu. (French) El a (*probabil) venit probabil. (Romanian) “He probably came” Il a bien mangé. (French) El a (*bine) mâncat bine. (Romanian) “He ate well” Les enfants ont tous vu (*tous) de(s) bons films. (French) Copiii au (*toți) văzut toți filme bune (Romanian) “The children have all seen good movies”

It is also important to mention that the Latin-to-Romance transition is characterised by a mixture between a transitional ‘broad’ V2 grammar4 4

Ledgeway (2007: 122–123) introduces the distinction between strict V2 grammars (the Germanic type) and broad V2 grammars (the medieval Romance type); the medieval Romance V2 features V-to-C movement but no ‘bottleneck effect’ (Poletto 2002), hence “whether the verb superficially occurs, for example,

92

Chapter Four

(Ledgeway 2008; 2012; cf. Benincà 1983; Adams 1987), mainly specific to root clauses, and an emerging V-to-I grammar, characterising embedded clauses (Ledgeway 2012). A prediction of this line of reasoning is that variation in the level of verb movement of the type low verb movement vs V-to-I movement should occur especially in embedded clauses where V-to-C movement (a syntactic option of old Romance) is a priori blocked by the lexicalization of C-elements (Ledgeway mss). This prediction is borne out in Martins’ (2002) analysis of old Portuguese, where it is reported that scrambling (alongside with the OV order) is mostly attested in subordinate clauses; the analysis herein has similar results (§2.2.2.2). 2.1.2 The pronominal clitic orientation parameter There is general agreement on the idea that clitics adjoin to the edge of the three functional domains of the clause in (7) above (Benincà and Tortora 2010; cf. also Rivero 1997; Roberts 2010; Tortora 2014b), this hypothesis giving rise to a three-way typology of pronominal clitic orientation:5 C-oriented vs I-oriented vs v/V-oriented clitics. Clitic orientation is language-specific (Rivero 1997), even construction-specific (Tortora 2014b). In order to distinguish C-oriented cliticization from the other two sites of cliticization, Rivero (1997) and subsequent research proposed the following diagnostics: the position of the clitic with respect to clausal negation and adjacency relations. C-oriented clitics precede clausal negation and are adjacent to complementisers; by contrast, clitics adjoining lower are preceded by clausal negation and establish adjacency relations with the verb. The examples below show that Romanian clitics are I-oriented: clausal negation systematically precedes the clitic and not vice versa (12); the clitic is adjacent to the verb(al complex), not the complementiser (13).6 in first, second, third, or fourth position, the V2 generalization nonetheless consistently holds, insofar as the verb is invariably assumed to have moved to the vacant C position” (Ledgeway 2007: 122). 5 Older research (e.g. Rivero 1997) generally distinguishes between C- vs I-orientation of cliticization; more recent research (e.g. Tortora 2014a; Ledgeway current volume) introduces the distinction between I-oriented cliticization and v/V-oriented cliticization. 6 Pronominal clitics may be phonologically adjacent to complementisers if no other element merges in the space between the C-head and the verbal cluster (i); the occurrence of the subject (13) or of any other left-peripheral constituent in-between the C-element and the verbal cluster shows that clitics adjoin to the verb (ii), testifying thus to their I-orientation. It is not clear if phonological encliticization to a C-head has syntactic effects too (Adam Ledgeway, p.c.).

Syntactic Archaisms Preserved in a Contemporary Romance Variety

(12)

(13)

93

dacă Dumnezeu nu te (*nu) if God not CL.ACC.2SG not ajută (Romanian) helps “if God doesn’t help you” a. dacă el l-a if he CL.ACC.M.3SG=AUX.PERF.3SG făcut bine make.PPLE well “if he did it well” b. *dacă-l el a if=CL.ACC.M.3SG he AUX.PERF.3SG făcut bine make.PPLE well

To sum up, (modern) Romanian pronominal clitics are I-oriented; the cliticization site is thus the edge of the IP-domain (Săvescu-Ciucivara 2011), sandwiched between the CP-field, NegP (Zanuttini 1997) and the TAM-projections in the IP-field. 2.1.3 Summary 1) In modern Romanian, V-raising is cross-paradigmatically high: synthetic verbs raise to the highest TAM-projection in the complex split-IP field; in analytic structures, the lexical verb also undergoes raising to the I-domain. 2) Modern Romanian pronominal clitics are I-oriented. In what follows, we set the data of ORom and (modern) IstroRom against the diagnostics established so far, with the goal of putting forward a formal analysis of interpolation and scrambling phenomena in these varieties.

(i)

(ii)

dacă-l vede if=CL.ACC.M.3SG see.PRES.3SG “if she sees him” dacă atunci / pe Ion if the DOM John l-a văzut CL.ACC.M.3SG=AUX.PERF.3SG see.PPLE “if she saw him then / if she saw John”

94

Chapter Four

2.2 Syntactic analysis: interpolation and scrambling as low verb movement The following syntactic analyses of interpolation and scrambling have been put forward in the literature: (a) C-orientation of pronominal clitics (Rivero 1997; Poole 2007 for old Spanish); (b) Aux-to-C movement (Mensching 2012 for old Romance); (c) low verb movement (Mensching 2012 for old Romance Aux-XP-Part structures; Ledgeway and Lombardi 2005 for interpolation in Cosentino; Nicolae 2015 for ORom interpolation). It is incumbent on us to determine what the proper syntactic analysis of the ORom and IstroRom discontiguous structures is. Our analysis will capitalize on two factors: (i) the parametric settings of ORom and IstroRom relevant for interpolation and scrambling, and (ii) the internal and external distributional properties of these structures. 2.2.1 Parametric features of old Romanian and Istro-Romanian 2.2.1.1 Old Romanian. The TAM system of ORom is very similar to that of modern Romanian (Zafiu 2016); the same analysis can be pursued (§2.1.1 above). Pronominal clitics are systematically I-oriented: the ordering [pronominal clitic-negator-V] is not attested in the corpus analyses of ORom (Nicolae and Niculescu 2015, 2016); in the presence of negation, pronominal proclisis is almost generalized (14); especially in translations from Old Church Slavonic7 (15a) (but not only, (cf. (15b)), there is 7

One of the anonymous reviewers pointed out the fact that examples like (15a) and (17) are from Church Slavonic translations and “display a word order that is foreign to old Romanian”, and recommended that we should dispense with them. We have also stressed the importance of using texts directly written in Romanian in the diachronic analysis (see §1.2 above). However, we have chosen to keep and use in our argumentation examples of this sort because we believe that their existence actually strengthens the hypothesis that, while the word order of major constituents may be influenced by a foreign word order pattern, the parametric settings of functional elements like pronominal clitics cannot be overridden by language contact. For example, the linearization [pronominal clitic-negator(-V)], attested in Old Church Slavonic texts like Codex Marianus (i) (Pancheva 2008), is never found in old Romanian; very rarely and only in translations, pronominal enclisis to the right of clausal negation (i.e. [negator-V-pronominal clitic]), also an option of Old Church Slavonic (ii), is attested (15); however, this option is compatible with an I- or v-oriented site for pronominal cliticization, parametric options also reflected by original old Romanian writings (see the main text).

Syntactic Archaisms Preserved in a Contemporary Romance Variety

95

pronominal enclisis on the verb is found in sentences with negation (which may reflect a low, vP-edge cliticization site); either proclitic or enclitic, clitics are adjacent to the verb(al complex), not the complementiser (16)– (17). (14)

(15)

(16)

(i)

(ii)

nimini de rudele meale nu nobody of relatives my not m-a grijit CL.ACC.1SG=AUX.PERF.3SG take.care.PPLE “none of my relatives took care of me” (DÎ.1591–600:VIII) a. să nu SĂSUBJ not bucure-mi-se enjoy.SUBJ.3SG=CL.DAT.1SG=CL.REFL.3SG “he should not enjoy me” (CP1.1577: 62r) b. n-au ridicatu-să not=AUX.PERF.3SG raise.PPLE=CL.REFL.3SG nimeni cu nici o pâră nobody with no squeal “nobody raised any squeal (= nobody squealed)” (AB 279, apud Todi 2001: 49) că cu direptate şi giudeţ adevărat that with justice and judgement right toate le aduseş all.F.PL CL.ACC.F.3PL bring.PS.2SG asupră de noi over of us “that with justice and right judgement you brought them all upon us” (DDL.1679: 122)

ašte sę bi ne CL.REFL.3SG be.COND.3SG not if rodilg čkg tg born.PPLE.M.SG man this “if this man had not been born” (Codex Marianus, Pancheva 2008) sę Ěko ne prěžde krgsti that not first wash.PAST.3SG CL.REFL.3SG prgvěe oběda before meal “that he did not first wash before the meal” (Codex Marianus, Pancheva 2008)

96

(17)

Chapter Four

Ferecaţi flămânzii acmu că happy.M.PL hungry.M.PL.DEF now that sătura-se-vor feel.full.INF=CL.REFL.3PL=AUX.FUT.3PL “The hungry people will be happy, because they will feel full” (CT.1560–1: 126r-v)

ORom V-to-C movement exclusively manifests as a “long-head movement” (= LHM) (Rivero 1993) phenomenon: displacement to the C-domain targets exclusively the lexical verb, which moves across the auxiliary/pronominal clitic+auxiliary, giving rise to V-(Cl) Aux structures (18). In contrast to other older Romance varieties (old French, old Italian, old Spanish; Rivero 1993; Roberts 1994), Aux-to-C is not attested in ORom (Nicolae 2015; Hill and Alboiu 2016). (18)

Pusu-ne-am şi put.PPLE=CL.REFL.1PL=AUX.PERF.1PL and degetele mai jos fingers.DEF more down “We have put our fingers below” (DÎ.1579‒80:VI)

To sum up, we may safely exclude the C-orientation of clitics and Aux-to-C movement as potential sources of ORom interpolation and scrambling. 2.2.1.2 Istro-Romanian. The TAM system of IstroRom is similar to that of (Daco-)Romanian, yet less rich in certain respects. The analytic structures display an internal structure similar to their old and modern Romanian counterparts: the analytic past tense has the [haveauxiliary+past participle] structure, and the analytic future and conditional are based on want-auxiliaries plus the infinitive (Caragiu Marioțeanu 1977: 223–225). In the verbal domain, Croatian influence manifests itself especially in the high frequency of V-Aux (i.e. LHM-like) structures (19a), also found in ORom (see §2.2.1.2 above). In Croatian, the V-Aux linearization is due to the Wackernagel status of the auxiliary, a fact which has been taken to extend to IstroRom as well (Giusti and Zegrean 2015). While authors like Giusti and Zegrean (2015) indicate the V-Aux ordering as the only option “when the sentence only consists in auxiliary and past participle” (p. 126), in the corpora we have examined we have also identified sentence-initial

Syntactic Archaisms Preserved in a Contemporary Romance Variety

97

auxiliaries (19b-c); hence, IstroRom auxiliaries do not strictly obey a Wackernagel constraint. (19)

a.

b.

c.

și murít-a ăm vói̯ sk̯ä and die.PPLE=AUX.PERF.3SG in war “and he died in the war” (HS: 96) Bire. Ŭam slujit good AUX.PERF.1SG serve.PPLE un cesar an emperor “Good. I served an emperor” (TC: 10) ‒ Veț cumpara cesta gal’ire? AUX.FUT.2PL buy.INF this hen “Will you buy this hen?” (TC: 84)

IstroRom pronominal clitics display much more freedom of ordering with respect to the lexical verb and the auxiliary than their old and modern Romanian counterparts. The following orderings are attested: (i) Cl-Aux: Cl-Aux-V and V-Cl-Aux structures (20); (ii) Aux-Cl-V (21); and (iii) Cl-V-Aux (22). (20)

(21)

(22)

Ce ti-ŭam io what CL.DAT.2SG=AUX.PERF.1SG I.NOM facut? do.PPLE “What did I do to you?” (TC: 8) b. Pus-l’-a mărle. put.PPLE=CL.DAT.3SG=AUX.PERF.3SG hands “He put his hands on him” (TC: 8) Se nu, voi te ucide. if not AUX.FUT.1SG CL.ACC.2SG kill.INF “If not, I will kill you” (TC: 90) Ontrat pus-a suddenly put.PPLE=AUX.PERF.3SG Magdalena ăn bărse și vo Magdalena.ACC in bag and CL.ACC.F.3SG legat-a. (TC: 98) tie.PPLE=AUX.PERF.3SG “Suddenly, he put Magdalena in the bag and tied her”

a.

98

Chapter Four

Only option (i) is also found in old (and modern) Romanian, while (ii) and (iii) are specific to IstroRom. This distribution indicates that IstroRom pronominal clitics are subject to a mixed parametric setting of the type identified by Rivero (1997) for old Spanish: option (i) points to the I-orientation of clitics, while option (iii) signals that pronominal clitics are C-oriented (the cliticization site is above V-Aux inversion, analysed as V-to-C movement), an option specific to (Serbo-)Croatian (Rivero 1997). Furthermore, option (ii) points towards the existence Aux-to-C movement, another option specific to IstroRom (and other older Romance varieties), but not to old (and modern) Romanian. To conclude, the higher degree of freedom in the placement of clitics and auxiliaries in IstroRom indicates that C-orientation of pronominal clitics and Aux-to-C movement cannot a priori be excluded as sources for interpolation. The next section will determine which syntactic analysis properly accounts for IstroRom and ORom interpolation and scrambling. 2.2.2 The syntax of discontiguous structures In this section, we submit the ORom and IstroRom discontiguous structures to a set of syntactic diagnostics which indicates that this phenomenon can be analysed in a unitary fashion in both varieties: the lexical verb (the synthetic verb in interpolation (type A); the lexical verb in scrambling (type B)) does not undergo V-to-I movement—hence being confined to the vP domain; XPs interposed between the Aux / Cl / Cl+Aux and the lexical verb either occupy their base-generation position or are accommodated by Belletti’s (2004) “low IP-area”, i.e. the periphery of the phasal head v* (Chomsky 2001; 2008). 2.2.2.1 Internal properties. Interpolated/scrambled XPs are systematically bordered, to their left, by a functional element—Aux (23)-(24)/Cl (25)-(26)—or a cluster of functional elements—Cl–Aux (27)-(28)—and, to their right, by the lexical verb: (23)

eu încă am pre el I already AUX.PERF.1SG DOM him botezat (ORom) baptize.PPLE “I have already baptized him” (CM.1567: 258r)

Syntactic Archaisms Preserved in a Contemporary Romance Variety

(24)

(25)

(26)

(27)

(28)

99

io n-am I not=AUX.PERF.1SG bíre-nțeles (IstroRom) well-understand.PPLE “I did not understand (it) well” (SF: 168) să ne prentru eale SUBJ CL.ACC.1PL for them ispăsim (ORom) atone.SUBJ.1PL “that we should atone for them” (CCat.1560: 5r) Me tu conoști? (IstroRom) CL.ACC.1SG you.SG.NOM know.PRES.2SG “Do you know me?” (TC: 8) mă vor cu pietri CL.ACC.1SG AUX.FUT.3PL with stones împroșca (ORom) splatter.INF “they will throw stones at me” (PO.1582: 235) tu t-er you.SG.NOM CL.REFL.2SG=AUX.FUT.2SG dupa me fil’e ănsura (IstroRom) after my daughter marry.INF “You will marry my daughter” (TC: 14)

The interpolation/scrambling position is not specialized; the XP may be: a subject (29)-(30), another argument (DO in (23) above, IO in (31)-(32) below), and adjuncts (a PP in (27) above, an adverbial in (35)-(39) below). (29)

(30)

cum au şi Hristos iubit how AUX.PERF also Christ love.PPLE sfânta besearecă (ORom) holy.DEF church “how Christ also loved the holy church” (CM.1567: 261r) Cănd a cesaru when AUX.PERF.3SG emperor.NOM dozneit (IstroRom) find.out.PPLE “when the emperor found out” (TC: 23)

100

(31)

(32)

Chapter Four

carele am mie which AUX.PERF.1SG me.DAT săpat (ORom) dig.PPLE “which I have dug for myself” (PO.1582: 177) morta s-a death.NOM CL.REFL.3SG=AUX.PERF.3SG lu Martin rugat (IstroRom) DAT Martin pray.PPLE “death prayed to Martin” (TC: 17) Multiple XPs may also be interpolated:

(33)

(34)

prince ţie au Domnedzeu by-what you.DAT AUX.PERF.3SG God toate aceastea arătat (ORom) all these show.PPLE “how God showed you all these things” (PO.1582: 143) c-ŭaț voi mire because=AUX.PERF.2PL you.PL.NOM me.ACC și me fil’ privarit (IstroRom) and my son cheat.PPLE “because you cheated me and my son” (TC: 9)

Recall from §2.2.1.2 above that the mixed parametric settings of IstroRom pronominal clitics and auxiliaries leave open the possibility that interpolation in IstroRom (but not in ORom) may also result from Aux-toC movement. However, in IstroRom examples like (28) or (32) featuring interpolation we observe that the only linearization is Cl-Aux. Actually, in an extensive corpus analysis of the relation between the variation between Cl-Aux/Aux-Cl and interpolation and scrambling in IstroRom (Dragomirescu and Nicolae 2016), we show that the Aux-Cl linearization is never available with discontiguous structures. These considerations rule out the possibility that interpolation/scrambling may result from Aux-to-C movement. Let us now turn to the syntactic diagnostics that verify the hypothesis that discontiguous structures are based on a syntax involving low verb movement. First of all, the lexical verb occurs to the right of adverbials from the high (35)-(36) and low (37)-(38) adverb spaces, and of adverbials of manner (39) (cf. Cinque 1999, Schifano 2015b):

Syntactic Archaisms Preserved in a Contemporary Romance Variety

(35)

(36)

(37)

(38)

(39)

101

că vei acmuşu muri (ORom) that AUX.FUT.2SG now die.INF “that you will now die” (FT.1571−5: 1v) Tu-i sigurno pročiteit you=AUX.PERF.2SG certainly read čuda libri. (IstroRom) many books “You have certainly read many books” (apud Zegrean 2012) ce se pururea pomeneaşte (ORom) which CL.REFL.3SG always mention.PRES.3SG “which gets always mentioned” (Prav.1581: 258r) che n-a zaino vezut that not=AUX.PERF.3SG immediately see.PPLE soldatu (IstroRom) soldier.ACC “that he did not immediately see the soldier” (TC: 21) a. scutat-a cum listen.PPLE=AUX.PERF.3SG how a cela pul’ mușat AUX.PERF.3SG that birdie beautifully căntat (IstroRom) sing.PPLE “she listened to that nestling singing beautifully” (TC: 43) b. io n-am I not=AUX.PERF.1SG bíre-nțeles (IstroRom) well-understand.PPLE “I did not understand (it) well” (SF: 168)

Secondly, the lexical verb occurs to the right of subject floating Qs (and floating reflexives, (41b)): (40)

cându are fi toți mâncându when AUX.COND.3PL be.IRR all eat.GER din mâncările păgânilor (ORom) of dishes.DEF pagans.DEF.GEN “when they would all be eating from the pagans’ dishes” (DVT.1679-99: 318r)

102

(41)

Chapter Four

a.

b.

Și o zi iel’ and one day they.NOM s-av toți CL.REFL.3PL=AUX.PERF.3PL all.M.PL pus scupa ăn se start.PPLE together in their seliște (...) (IstroRom) village “And one day they all started together in their village…” (TC: 149) Ușai se va ănsai door CL.REFL.3SG AUX.FUT.3SG itself rescl’ide open.INF “The door will open by itself” (TC: 25, 26, 27)

Consider also example (42) from IstroRom which features an interpolated wh-element; the element sandwiched between the reflexive clitic and the lexical verb is the interrogative adverbial cum (‘how’), whose surface position reflects either its base-generated position (wh-in-situ) or a(n intermediate) landing site on the v*-edge. Whatever the exact position of cum (‘how’), we can be sure that the lexical verb has not undergone V-to-I movement in (42). (42)

La voi se cum zíče? (IstroRom) at you.PL CL.REFL.3SG how say.PRES.3SG “How do you say it in your language?” (SF: 72)

Finally, both ORom and IstroRom present Aux-licensed VP-ellipsis (43)-(44a) and pseudogapping (44b), elliptical phenomena specific to English-type languages displaying low verb movement (Giusti and Zegrean 2015 for IstroRom, Nicolae 2015 for ORom). The clitic nature of the auxiliary verb may be superseded by co-licensing (licensing by multiple auxiliaries) (43). (43)

de-l va fi if=CL.ACC.M.3SG AUX.FUT.3SG AUX.IRR mutat sau de nu-l moved.PPLE or if not=CL.ACC.M.3SG va fi √ (ORom) AUX.FUT.3SG AUX.IRR “whether he will have moved him or not” (Prav.1646: 78; Stan 2013)

Syntactic Archaisms Preserved in a Contemporary Romance Variety

(44)

a.

b.

103

‒ Ŭai dus? (IstroRom) AUX.PERF.2SG go.PPLE ‒ Ŭam √. AUX.PERF.1SG “‒ Did you go? ‒ I did” (TC: 8) ‒ Cum ŭai sta nopte how AUX.PERF.2SG this night durmit? sleep.PPLE ‒ Ŭam √ bire. AUX.PERF.1SG well “‒ How did you sleep this night?/ I have [slept] well.” (TC: 57)

Another important piece of evidence in favour of the hypothesis that discontiguity does not derive from a syntax based on V(Aux)-to-C movement or from the C-orientation of pronominal clitics in the varieties we are investigating comes from the fact that these phenomena are also attested in negative clauses. It is well-known that clausal negators block Vto-C movement in the Romance languages (Pollock 1989; Rivero 1993; Roberts 1994; Isac and Jakab 2004); the prediction of this line of reasoning is that scrambling would not be available in negative clauses if its syntax were based on Aux-to-C movement, contrary to fact as shown in (45), (46). (45)

(46)

deaca nu va omul pre if not AUX.FUT.3SG man.DEF on ceastă lume în viiața sa, purta this world in life.DEF his bear.INF grije (ORom) worry “if man were not worried in this world, in his life” (CC1.1567: 241r) N-åu de månt’’e lucråt (IstroRom) not=AUX.PERF.3SG before work.PPLE “He didn’t work before” (SF: 86)

Similarly, interpolation is available in negative clauses, with pronominal clitics occurring to the right of negation (hence, being IP-oriented) (47), (48).

104

(47)

(48)

Chapter Four

și nu-șu pre niminea and not=CL.REFL.3SG DOM nobody prepunea elu, numai pre suspected he only DOM păcatele lui (ORom) sins.DEF his “he wasn’t suspecting anyone, just his sins” (CC2.1581: 57) Nu se nícad iåvę (IstroRom) not CL.REFL.3SG never showed “He never showed himself” (SF, 80)

These observations neatly correlate with the results from Dragomirescu and Nicolae (2016) (reported above) where it is shown that the Cl-Aux ordering (an option of IstroRom) is actually never available in the presence of scrambling. Taking stock of all the properties investigated above, we may conclude that, in structures with interpolation/scrambling, the lexical verb is placed in the vP-domain and does not undergo raising to the I-domain. One of the reviewers argues against the low verb movement analysis proposed here on the basis of ORom examples like (49) below (also available in IstroRom, (50)), in which postverbal subjects are available in structures with scrambling: (49)

(50)

până vor mai bine înţeleage until AUX.FUT.3PL more well understand.INF creştinii (ORom) Christians.DEF “until the Christians will have a better understanding” (CM.1567: 263r) De månt’’e s-au la noi before CL.REFL.3PL=AUX.PERF.3PL at us așå purtåt ómeri (IstroRom) like.this behave.PPLE people “people have always behaved like this at us (= in our community)” (SF: 58)

We believe that examples of this sort are not counterarguments to the proposed analysis. To begin with, witness in (49) the position of the manner adverbial (mai) bine (“better”): ever since Cinque (1999), it has been largely accepted that manner adverbs mark the boundary between the IP-domain and the lowest clausal domain. Secondly and more importantly,

Syntactic Archaisms Preserved in a Contemporary Romance Variety

105

under the dynamic phase approach (e.g. den Dikken 2007, Gallego 2010, Harwood 2015) adopted here, the clause internal phase labelled as the v*-phase contains more material than the minimal vP. As explained in §2.1.1 above, there are authors who consider that the low phasal domain extends up to Voice (with VoiceP > vP) (Baltin 2012; Rouveret 2012; Roberts 2013), and even authors who extend the phasehood of the clause internal phase up to (low) Aspect (Harwood 2015)—theoretical analyses like Gallego’s (2010) “phase sliding” or den Dikken’s (2007) “phase extension” naturally permit the extension of the low clausal domain. What is however important is that the verb is situated on the edge (Voice/(low)Asp) of the vP-phase in order to satisfy the Phase Impenetrability Condition (Chomsky 2001) and is accessible for probing by the IP-related heads. One remaining point of divergence between ORom and IstroRom is the mixed parametric setting of pronominal clitics in IstroRom. The Corientation of clitics in IstroRom—resulting from language contact with Croatian—may be one of the sources of interpolation in this idiom. The fact of the matter is that in interpolation without auxiliaries, pronominal clitics display a genuinely P(arametrically)-ambiguous grammar (Clark and Roberts 1993), being compatible with a grammar with both C-orientation and I-orientation. 2.2.2.2 External properties. In previous works on interpolation and scrambling it has been noticed that these phenomena overwhelmingly occur in embedded clauses (old Portuguese, Martins 2002; old Spanish, Poole 2007). The very same is also true of interpolation and scrambling in ORom and IstroRom. The table below captures the numerical distribution of discontiguous structures in main vs embedded clauses in a few representative texts from our ORom and IstroRom corpora. Since interpolation is extremely rare, our statistics is based on scrambling (i.e. with Aux/Cl-Aux structures).

106

Chapter Four

Table 1: Interpolation in main vs embedded clauses Text ORom

DÎ Texte PO.1582 Total ORom

IstroRom

TC AD AK Total IstroRom

Total number 5 13 29 47 (100%) 933 105 92 1130 (100%)

Main clause 2 3 3 8 (17%) 451 27 39 517 (46%)

Embedded clause 3 10 26 39 (83%) 482 78 53 613 (54%)

Recall from §2.1.1 above that the Latin-to-Romance transition is characterized by two phenomena in the area of V-raising: a transitional ‘broad’ V2 grammar, mainly specific to main clauses, and an emerging V-to-I grammar, characterising embedded clauses (where V-to-C is generally blocked by the lexicalization of C-elements, Ledgeway 2012; mss). Importantly, in main clauses, interpolation occurs mostly in whinterrogatives; this is, once more, a situation in which V-to-C movement is bled by wh-movement to the CP-domain. It has been observed that in both ORom (Hill and Alboiu 2016) and IstroRom (Dragomirescu and Nicolae 2016) V-to-C movement (irrespective of its trigger) and wh-movement are generally in complementary distribution. The predominant occurrence of interpolation and scrambling—analysed here as involving low verb movement—in clauses in which V-to-C movement is potentially blocked represents an important (yet often overlooked) piece of evidence for gradual emergence of V-to-I movement in the Latin-to-Romance transition. 2.2.2.3 Convergence and divergence in Istro-Romanian and Croatian. The preservation of interpolation and scrambling in IstroRom is atypical from a Romance comparative perspective. Naturally, intensive Istro-Romanian/Croatian bilingualism probably played a role in the preservation of these archaic Romance phenomena in IstroRom. The parallel Croatian/IstroRom texts in Kovačec (1998) allow us to analyse, at least partially, the degree to which Croatian influenced IstroRom with respect to interpolation and scrambling. The exhaustive analysis of Kovačec’s corpus (AK) has revealed four situations:

Syntactic Archaisms Preserved in a Contemporary Romance Variety

107

1) Discontiguity is found only in IstroRom; the verbal form in Croatian is synthetic: (51)

a.

b.

cum mi-ai how CL.DAT.1SG=AUX.PERF.2SG tu zis (IstroRom, AK: 324) you.SG.NOM say.PPLE kao što mi ti how what me you rekla (Croatian, AK: 325) say.PAST “how you said (this) to me”

2) Discontiguity is found only in IstroRom; the verbal form in Croatian is also analytic, but does not display interpolation: (52)

a.

b.

cân a fare when AUX.PERF.3SG outside iesít (IstroRom, AK: 323) go.out.PPLE kada su izašle when is.AUX go.out.PAST van (Croatian, AK: 324) outside “when he went out”

3) Discontiguity is found in both IstroRom and Croatian with the same interpolated constituent: (53)

a.

b.

ke s-a that CL.REFL.3SG=AUX.PERF.3SG cosíțele pletít (IstroRom, AK: 327) ponytails braid.PPLE si je pletenice CL.REFL.3SG is.AUX ponytails plela (Croatian, AK: 328) braid.PAST “that she braid her ponytails”

4) Discontiguity is found in both IstroRom and Croatian, but the interpolated constituent is not the same:

108

(54)

Chapter Four

a.

b.

e ie zis-a and he say.PPLE=AUX.PERF.3SG ke-l’ va cu sabl’a that=CL.DAT.3SG AUX.FUT.3SG with sword capu tal’a (IstroRom, AK: 328) head cut.INF a on je rekao da and he is.AUX say.PAST that ce joj sabliom odsjeći AUX.FUT.3SG her.DAT with.sword cut glavu (Croatian, AK: 328) head “and he said that he will cut her head with the sword”

The four situations identified above indicate that, while Croatian influences IstroRom to a certain degree, IstroRom does not perfectly mimic Croatian word order. Rather, as often happens in language contact, overlapping options consolidate features of the language being influenced which elsewhere disappear (Heine and Kuteva 2005: 50 and passim).

3. Conclusions: interpolation and scrambling in a broader diachronic and comparative perspective 1) The ORom and IstroRom data may be analysed in a uniform fashion: in both idioms, interpolation and scrambling result from the fact that V-to-I raising of the lexical verb does not take place (i.e. low verb movement). 2) Interpolation and scrambling may be added to the list of phenomena found in ORom which became extinct in the passage to modern Romanian,8 but were preserved in IstroRom (e.g. V-(Cl-)Aux structures, §§2.2.1.1, 2.2.1.2 above; auxiliary omission, non-clitic doubled and nondifferentially marked DOs, analytic datives introduced by the preposition A, etc.—see Dragomirescu and Nicolae 2016 for the inventory of ORom/IstroRom convergent phenomena). A common old Romance and 8

In modern Romanian, the verbal cluster may be broken only by the clitic adverbials mai, cam, prea, tot, and și. However, these elements have been analysed as heads (X0-elements) which incorporate into the structure of the lexical verb (see Giurgea 2011; Mîrzea Vasile 2015; Nicolae 2015). Therefore, they are not of interest for interpolation/scrambling, phenomena which involve the interposition of XPs.

Syntactic Archaisms Preserved in a Contemporary Romance Variety

109

proto-Romanian ancestry may be invoked for the existence of interpolation and scrambling in both ORom and IstroRom. However, mixed Romance and Slavic (i.e. Croatian) parametric options have favoured the preservation of interpolation and scrambling, as well as of other syntactic archaisms, in a modern isolated speech community, IstroRom. This indicates that, in the domain of syntax, one of the effects of language contact is the consolidation of archaic features which otherwise become extinct. 3) From a diachronic Romance perspective, the existence of low verb movement phenomena in syntactic contexts in which V-to-C movement is potentially blocked (e.g. embedded clauses; main clauses with whmovement) indicates that the Latin-to-Romance transition in the domain of verb movement is actually characterised by two distinct transitional processes: a ‘broad’ V2 grammar, specific especially to main clauses in which V-to-C movement is potentially free, and a gradual transition from a grammar with low verb movement to a grammar with V-to-I movement, found in contexts in which V-to-C movement is potentially restricted, i.e. in embedded clauses (by the lexicalization of C-elements) or in main clauses with wh-movement.

Acknowledgements We would like to express our gratitude to the audiences of the Stony Brook Linguistic Symposium on Romance Languages (March 31–April 3, 2016) and of the Bucharest Formal Approaches to Romance Microvariation workshop (November 25–26, 2016) where parts of the paper were presented, and to Adam Ledgeway, Martin Maiden, Ionuț Geană and the two anonymous reviewers who read our manuscript, made valuable suggestions and asked important questions which helped improve it. Remaining errors are, of course, ours.

Old Romanian corpus AB = Istoria Țării Rumânești de la octombrie 1688 pînă la martie 1717, ed. by Mihail Gregorian, Cronicarii munteni, II, 273-352. (apud Todi 2001) BB.1688 = Biblia. Ed.: Biblia adecă Dumnezeiasca Scriptură a Vechiului şi Noului Testament, tipărită întâia oară la 1688 în timpul lui Şerban Vodă Cantacuzino, Domnul Ţării Româneşti, Bucharest, Editura Institutului Biblic, 1977. CC1.1567 = Coresi, Tâlcul Evangheliilor. Ed.: Coresi, Tâlcul evangheliilor şi molitvenic românesc, ed. V. Drimba, Bucharest, Editura Academiei Române, 1998, 31–187.

110

Chapter Four

CCat.1560 = CCat.1560 = Coresi, Catehism. Ed. Al. Roman-Moraru, in: Texte, 101–105. CM.1567 = Coresi, Molitvenic. Ed.: Coresi, Tâlcul evangheliilor şi molitvenic românesc, ed. Vladimir Drimba, Bucharest, Editura Academiei Române, 1998, 189–211. CP1.1577 = Coresi, Psaltire slavo-română. Ed.: Coresi, Psaltirea slavo-română (1577) în comparaţie cu psaltirile coresiene din 1570 şi din 1589, ed. Stela Toma, Bucharest, Editura Academiei, 1976, 35–662. CT.1560–1 = Coresi, Tetraevanghel. Ed.: Tetraevanghelul tipărit de Coresi. Braşov 1560–1561, comparat cu Evangheliarul lui Radu de la Măniceşti. 1574, ed. Florica Dimitrescu, Bucharest, Editura Academiei, 1963. DDL.1679 = Dosoftei, Dumnezăiasca liturghie. Ed. N. A. Ursu, Iași, Mitropolia Moldovei și Sucevei, 1980, 3–313. DÎ = Documente şi însemnări româneşti din secolul al XVI-lea, text stabilit şi indice de Gheorghe Chivu, Magdalena Georgescu, Magdalena Ioniţă, Alexandru Mareş, Alexandra Roman-Moraru, Bucharest, Editura Academiei, 1979. DVT.1679-99 = O traducere inedită a Vechiului Testament din secolul al XVI-lea. Ed. C. Dima, Bucharest, Editura Universității din București, 2009, 110–217. FT.1571−5 = Fragmentul Todorescu (Carte de cântece). Ed. Ion Gheţie, in: Texte, 336–343. PO.1582 = Palia de la Orăştie. Ed. V. Pamfil, Bucharest, Editura Academiei, 1968. Prav.1581 = Pravila ritorului Lucaci. Ed. I. Rizescu, Bucharest, Editura Academiei, 1971, 161–183. Prav.1646 = Carte românească de învăţătură. Ed.: Carte românească de învăţătură. 1646, ed. Colectivul pentru vechiul drept românesc condus de acad. Andrei Rădulescu, Bucharest: Editura Academiei, 1961, 33–106. Texte = Ion Gheţie (coord.), Texte româneşti din secolul al XVI-lea. I. Catehismul lui Coresi; II. Pravila lui Coresi; III. Fragmentul Todorescu; IV. Glosele Bogdan; V. Prefeţe şi Epiloguri, Bucharest, Editura Academiei, 1982, 101– 105.

Istro-Romanian corpus AD = Dianich, Antonio, Vocabolario istroromeno-italiano. La varietà istroromena di Briani (’Bǝršćina), Pisa, Edizioni ETS, 2010. AK = Kovačec, August, Istrorumunjsko-hrvatski rječnik (s gramaticom i tekstovima), Pula, 1998. HS = „Harta sonoră” a graiurilor și a dialectelor limbii române, ed. by Maria Marin, Marilena Tiugan, Bucharest, Editura Academiei Române, 2014. SF = Sârbu, Richard, Vasile Frățilă, Dialectul istroromân. Texte și glosar, Timișoara, Editura Amarcord, 1998. TC = Texte istroromâne culese de Traian Cantemir, Bucharest, Editura Academiei, 1959.

Syntactic Archaisms Preserved in a Contemporary Romance Variety

111

References Adams, M., 1987, Old French, Null Subjects, and Verb Second Phenomena, PhD dissertation, UCLA. Alboiu, G., V. Motapanyane, 2000, “The generative approach to Romanian grammar: an overview”, in: V. Motapanyane (ed.), Comparative Studies in Romanian Syntax, Amsterdam, Elsevier, 1–48. Avram, M., [1975] 2007, “Particularităţi sintactice neromâneşti în diferite momente ale evoluţiei limbii române literare”, in: Studii de sintaxă a limbii române, Bucharest, Editura Academiei Române, 93–103. Baltin, M., 2012, “Deletion versus pro-forms: an overly simple dichotomy?”, Natural Language & Linguistic Theory, 30, 2, 381–423. Belletti, A., 2004, “Aspects of the low IP area”, in: L. Rizzi (ed), Structure of CP and IP. The Cartography of Syntactic Structures, vol. 2, Oxford, Oxford University Press, 16–51. Benincà, P., 1983, “Un’ipotesi sulla sintassi delle lingue romanze medievali”, Quaderni Patavini Di Linguistica, 4, 3–19. Benincà, P., C. Tortora, 2010, “On clausal architecture: evidence from complement clitic placement in Romance”, in: V. Torrens, L. Escobar, A. Gavarró, J. Gutiérrez (eds), Movement and Clitics: Adult and Child Grammar, Newcastle, Cambridge Scholars Publishing, 219–237. Biberauer, T., I. Roberts, 2010, “Subjects, tense, and verb movement”, in: T. Biberauer, A. Holmberg, I. Roberts, M. Sheehan (eds), Parametric variation: Null subjects in minimalist theory, Cambridge, Cambridge University Press, 263–302. Caragiu Marioțeanu, M., 1977, “Dialectul istroromân”, in: M. Caragiu Marioțeanu, Ș. Giosu, L. Ionescu-Ruxăndoiu, R. Todoran, Dialectologie română, Bucharest, Editura Didactică și Pedagogică, 213‒230. Chomsky, N., 2001, “Derivation by phase”, in M. Kenstowicz (ed.), Ken Hale: A Life in Language, Cambridge, MA, MIT Press, 1–52. Chomsky, N., 2008, “On phases”, in: R. Freidin, C. P. Otero, M. L. Zubizarreta (eds), Foundational Issues in Linguistic Theory. Essays in Honor of JeanRoger Vergnaud, Cambridge, MA, MIT Press, 133–166. Cinque, G., 1999, Adverbs and Functional Heads, Oxford, Oxford University Press. Clark, R., I. Roberts, 1993, “A computational approach to language learnability and language change”, Linguistic Inquiry, 24, 2, 299–345. Coteanu, I., 1957, Cum dispare o limbă: istroromîna, Bucharest, Societatea de Științe Istorice și Filologice. den Besten, H., 1983, “On the interaction of root transformations and lexical deletive rules”, in: W. Abraham (ed.), On the Formal Syntax of the Westgermania, Amsterdam, John Benjamins, 47–61. den Dikken, M., 2007, “Phase Extension Contours of a theory of the role of head movement in phrasal extraction”, Theoretical Linguistics, 33, 1, 1–41. Dobrovie-Sorin, C., 1994, The Syntax of Romanian, Berlin, Mouton de Gruyter.

112

Chapter Four

Dragomirescu, A., 2013, “O schimbare parametrică de la româna veche la româna modernă în sintaxa formelor verbale compuse cu auxiliar”, Limba română, 62, 2, 225–239. Dragomirescu, A., 2015, “Există trăsături slavone în sintaxa limbii române? Două studii de caz / Are there Slavonic features in the syntax of Romanian? Two case studies”, Diacronia 1, 1, doi:10.17684/i1A3ro; doi:10.17684/i1A3en. Dragomirescu, A., A. Nicolae, 2016, “O trăsătură sintactică a românei vechi păstrată în istroromână: interpolarea”, Limba română, 65, 4, 454–464. Dragoş, E., 1995, Elemente de sintaxă istorică românească, Bucharest, Editura Didactică şi Pedagogică. Emonds, J., 1978, “The verbal complex V’-V in French”, Linguistic Inquiry, 9, 2, 151–175. Fischer, S., 2003, The Catalan Clitic System. A Diachronic Perspective on its Syntax and Phonology, Berlin, Mouton de Gruyter. Gallego, J. Á., 2010, Phase Theory, Amsterdam, John Benjamins. Gheţie, I., 1975, Baza dialectală a românei literare, Bucharest, Editura Academiei. Giurgea, I., 2011, “The Romanian verbal cluster and the theory of head movement”, in: J. Herschensohn (ed.), Romance linguistics 2010. Selected Papers from the 40th Linguistic Symposium on Romance Linguistics (LSRL), Seattle, Washington, March 2010, Amsterdam, John Benjamins, 271–286. Giusti, G., I. Zegrean, 2015, “Syntactic protocols to enhance inclusive cultural identity. A case study on Istro-Romanian clausal structure”, Quadermi di linguistica e studi orientali/Working papers in linguistics and oriental studies, 1, 117‒138. Harwood, W., 2015, “Being progressive is just a phase: celebrating the uniqueness of progressive aspect under a phase-based analysis”, Natural Language & Linguistic Theory, 33, 2, 523–574. Heine, B., T. Kuteva, 2005, Language Contact and Grammatical Change, Cambridge, Cambridge University Press. Hill, V., G. Alboiu, 2016, Verb Movement and Clause Structure in Old Romanian, Oxford, Oxford University Press. Isac, D., E. Jakab, 2004, “Mood and Force features in the languages of the Balkans”, in: O. Mišeska Tomić (ed.), Balkan Syntax and Semantics, Amsterdam, John Benjamins, 315–338. Koster, J., 1975, “Dutch as an SOV Language”, Linguistic Analysis, 1, 111–136. Kovačec, A., 1971, Descrierea istroromânei actuale, Bucharest, Editura Academiei. Ledgeway, A., 2007, “Old Neapolitan word order: some initial observations”, in: A. L. Lepschy, A. Tosi (eds), Languages of Italy. Histories and Dictionaries, Ravenna, Longo Editore Ravenna, 119–146. Ledgeway, A., 2008, “Satisfying V2 in Early Romance: Merge vs. Move”, Journal of Linguistics, 44, 2, 437–470. Ledgeway, A., 2012, From Latin to Romance: Morphosyntactic Typology and Change, Oxford, Oxford University Press. Ledgeway, A., 2015, “Romance auxiliary selection in light of Romanian evidence”, in: G. Pană Dindelegan, R. Zafiu, A. Dragomirescu, I. Nicula,

Syntactic Archaisms Preserved in a Contemporary Romance Variety

113

A. Nicolae, L. Esher (eds), Diachronic Variation in Romanian, Newcastle upon Tyne, Cambridge Scholars Publishing, 3–31. Ledgeway, A., A. Lombardi, 2005, “Verb movement, adverbs, and clitic positions in Romance”, Probus, 17, 1, 77–101. Ledgeway, A., current volume, “Parallels in clausal and nominal structures: Romanian clitic placement”. Ledgeway, A., mss, “Late Latin verb second: the sentential word order of the Itinerarium Egeriae”, mss, University of Cambridge. Maiden, M., 2016, “Romanian, Istro-Romanian, Megleno-Romanian, and Aromanian”, in: A. Ledgeway, M. Maiden (eds.), The Oxford Guide to the Romance Languages, Oxford, Oxford University Press, 91–125. Martins, A. M., 2002, “The loss of IP-scrambling in Portuguese: clause structure, word-order variation and change”, in: D. Lightfoot (ed.), Syntactic Effects of Morphological Change, Oxford, Oxford University Press, 232–248. Mensching, G., 2012, “Parameters in Old Romance word order: a comparative minimalist analysis”, in: C. Galves, S. Cyrino, R. Lopes, F. Sandalo, J. Avelar (eds), Parameter Theory and Language Change, Oxford, Oxford University Press, 21–42. Mîrzea Vasile, C., 2015, “The position of the light adverbials și, cam, mai, prea, and tot in the verbal cluster: synchronic variation and diachronic observations”, in: G. Pană Dindelegan, R. Zafiu, A. Dragomirescu, I. Nicula, A. Nicolae, L. Esher (eds), Diachronic Variation in Romanian, Newcastle upon Tyne, Cambridge Scholars Publishing, 385–416. Nicolae, A., 2015, Ordinea constituenţilor în limba română. O perspectivă diacronică. Structura propoziţiei şi deplasarea verbului, Bucharest, Editura Universităţii din Bucureşti. Nicolae, A., D. Niculescu, 2015, “Pronominal clitics in old Romanian: the ToblerMussafia Law”, Revue roumaine de linguistique, 60, 2–3, 223–242. Nicolae, A., D. Niculescu, 2016, “Pronominal clitics: clitic ordering, clitic clusters”, in: G. Pană Dindelegan (ed.), The Syntax of Old Romanian, Oxford, Oxford University Press, 52–70. Pană Dindelegan, G., 2013, “Romanian—a brief presentation”, in: G. Pană Dindelegan (ed.), The Grammar of Romanian, Oxford, Oxford University Press, 1–7. Pană Dindelegan, G., A. Dragomirescu, 2016, “Conclusions”. in: G. Pană Dindelegan (ed.), The Syntax of Old Romanian, Oxford, Oxford University Press, 629–637. Pancheva, R., 2008, “Head-directionality of TP in Old Church Slavonic”, in: A. Antonenko, J. Bailyin, and C. Bethin (eds.), Formal Approaches to Slavic Linguistics 16: the Stony Brook meeting, 2007, Ann Arbor, Michigan Slavic Publications, 313–332. Poletto, C., 2002, “The left periphery of a V2-Rhaetoromance dialect: a new perspective on V2 and V3”, in: S. Barbiers, L. Cornips, S. van der Kleij (eds), Syntactic Microvariation. Proceedings of the Workshop on Syntactic Microvariation, Amsterdam, August 2000, Amsterdam, Meertens Institute, 214–242.

114

Chapter Four

Poletto, C., 2014, Word Order in Old Italian, Oxford, Oxford University Press. Pollock, J.-Y., 1989, “Verb movement, Universal Grammar, and the structure of IP”, Linguistic Inquiry, 20, 3, 365-424. Poole, G., 2007, “Interpolation and the left periphery in Old Spanish”, in: M. Hussein, M. Kolokante, C. Wright (eds), Newcastle Working Papers in Linguistics, 13, 188–216. Rivero, M.-L., 1993, “Long head movement vs. V2, and null subjects in old Romance”, Lingua, 89, 2–3, 217–245. Rivero, M.-L., 1997, “On two locations for complement-clitic pronouns: SerboCroatian, Bulgarian and Old Spanish”, in: A. van Kemenade, N. Vincent (eds), Parameters of Morphosyntactic Change, Cambridge, Cambridge University Press, 170–206. Roberts, I., 1994, “Two types of head movement in Romance”, in: D. Lightfoot, N. Hornstein (eds), Verb Movement, Cambridge, Cambridge University Press, 207–242. Roberts, I., 2010, Agreement and Head Movement. Clitics, Incorporation, and Defective Goals, Cambridge, MA, MIT Press. Roberts, I., 2013, “Some speculations on the development of the Romance periphrastic perfect”, Revue roumaine de linguistique, 58, 1, 3–30. Rouveret, A., 2012, “VP ellipsis, phases and the syntax of morphology”, Natural language & Linguistic Theory, 30, 3, 897–963. Salvi, G., 2001, “The two sentence structures of Early Romance”, in: G. Cinque, G. Salvi (eds), Current Studies in Italian Syntax. Essays offered to Lorenzo Renzi, Amsterdam—London—New York—Oxford—Paris—Shannon— Tokyo, Elsevier, 297–312. Salvi, G., 2004, La formazione della struttura di frase romanza. Ordine delle parole e clitici dal latino alle lingue romanze antiche, Tübingen, Niemeyer. Săvescu Ciucivara, O., 2011, A Syntactic Analysis of Pronominal Clitic Clusters in Romance. The View from Romanian, Bucharest, Editura Universităţii din Bucureşti. Schifano, N., 2015a, “The paradigmatic instantiation of TAM. A novel approach to Romance verb-movement”, in: E. O. Aboh (ed.), Romance Languages and Linguistic Theory 2013. Selected Papers from ‘Going Romance’ Amsterdam 2013, Amsterdam, John Benjamins, 83–102. Schifano, N., 2015b, Verb Movement: A Pan-Romance Investigation, PhD dissertation, University of Cambridge. Sitaridou, I., 2012, “A comparative study of word order in Old Romance”, Folia linguistica, 46, 2, 553–604. Stan, C., 2013, O sintaxă diacronică a limbii române vechi, Bucharest, Editura Universităţii din Bucureşti. Timotin, E., 2016, “Presenting the corpus: typologizing, dating, and locating the texts”, in: G. Pană Dindelegan (ed.), The Syntax of Old Romanian, Oxford, Oxford University Press, 1–7. Todi, A., 2001, Elemente de sintaxă românească veche, Pitești, Editura Paralela 45. Tortora, C., 2014a, A Comparative Grammar of Borgomanerese, Oxford, Oxford University Press.

Syntactic Archaisms Preserved in a Contemporary Romance Variety

115

Tortora, C., 2014b, “Clausal domains and clitic placement generalizations in Romance”, in: K. Lahousse, S. Marzo (eds), Romance Languages and Linguistic Theory 2012: Selected papers from 'Going Romance' Leuven 2012, Amsterdam, Benjamins, 1–36. Vasiliu, E., L. Ionescu-Ruxăndoiu, 1986, Limba română în secolele al XII-lea—al XV-lea (fonetică—fonologie − gramatică), Bucharest, Tipografia Universității din București. Zafiu, R., 2016, “The syntax of moods and tenses”, in: G. Pană Dindelegan (ed.), The Syntax of Old Romanian, Oxford, Oxford University Press, 14–52. Zanuttini, R., 1997, Negation and Clausal Structure. A Comparative Study of Romance Languages. Oxford, Oxford University Press. Zegrean, I. G., 2012, Balkan Romance: Aspects of the Syntax of Istro-Romanian, PhD dissertation, Università Ca’Foscari Venezia.

CHAPTER FIVE THE CLITIC DOUBLING CYCLE: A DIACHRONIC RECONSTRUCTION JORGE VEGA VILANOVA, SUSANN FISCHER, MARIO NAVARRO1 University of Hamburg [email protected]; [email protected]; [email protected]

1. Introduction Clitic Doubling (CLD) is generally understood as a construction in which a clitic co-occurs with a full DP in argument position forming a discontinuous constituent with it.2 (1)

doy el libro give.1SG the book María.3 (Stand. Sp.) Mary “I give him/Mary the book.” Le

CL.3SG

1

a to

él/a him/to

The authors would like to acknowledge the DFG grant FI 875/3-1 “Clitic Doubling Across Romance” for financial support of their research. 2 Many linguists have claimed that there is no structural difference between CLD configurations and clitic left and right dislocation structures (see e.g. Aoun 1981). However, Jaeggli (1986) argues that dislocated structures involve prosodic cues absent in CLD. Furthermore, clitic left and right dislocation allows wide and narrow scope whereas CLD only allows wide scope (cf. Torregrossa 2012). 3 In all examples, the clitics appear in bold letters and the accompanying DPs are underlined.

118

(2)

Chapter Five

El CL.3SG.ACC

vam veure saw.1PL

a ell. (Stand. Cat.) to him.

“We saw him.” The phenomenon of clitic doubling has always been of paramount interest among generative linguists, circling around the question of whether clitics move from an argument position to their surface position or whether they are base generated in the position they end up in, functioning there as some kind of agreement marker. The theoretically most interesting question has always been the fact that the clitic and DP seem to share one theta-role and one case. Over the years many different theories have been proposed in order to explain what conditions the distribution of clitic doubling synchronically, and what triggers the emergence of clitic doubling diachronically. So far, no generally accepted theory has been proposed. In this paper we will present and discuss data from Old Spanish (OS) and modern Spanish varieties (Buenos Aires, Lima, Andean Spanish, and Judeospanish) and data from Old Catalan (OC) and Modern Catalan (Barcelona). The empirical evidence we have collected allows the assumption that the emergence of CLD can be explained by a cyclic change, more specifically by a cycle that consists of five stages. Furthermore, we will argue that CLD is an epiphenomenon that emerges from the interaction of word order changes, and the grammaticalization process of the doubling DP and the clitic itself.

2. The phenomenon across languages Clitic doubling is a structure that can be found across different language families. It has been argued to exist within the Indo-European languages in Romance (e.g. in Spanish), Semitic (e.g. in Hebrew), Slavic (e.g. in Bulgarian, Macedonian), Albanian, and Greek. It was also argued to be a feature of Degema (Kari 2003) and Pirahā (Everett 1987). However, CLD is a highly variable phenomenon. On the one hand, not all languages within a language family allow this structure: e.g. within the Romance language family, CLD is allowed in Spanish (Strozer 1976; Rivas 1977), Romanian (Dobrovie-Sorin 1990), Catalan (Fischer 2002), and Portuguese (Quícoli 1976; Madeira 1993), but not in French (Kayne 19754) or Italian (Belletti 19995). On the other hand, also within a 4 5

But see Kayne (2000) for a different view on the spoken language. Contra Ledgeway (2000), Poletto (2000) and Pescarini (2015).

The Clitic Doubling Cycle: A Diachronic Reconstruction

119

language, different patterns of CLD are attested. For instance, CLD of the direct object (DO) is more restricted in Standard Spanish than it is in its varieties. In most Spanish varieties, it is obligatory when the DO is a stressed pronoun (e.g. Pedro *(lo) vio a él ‘Pedro saw him’), in others it is also allowed when the DO is a full DP (e.g. Buenos Aires Spanish Pedro (lo) llamó al mozo ‘Pedro called the waiter’). Moreover, for indirect objects (IOs), doubling is mostly optional (e.g. (Le) devolví el coche a Pedro ‘I gave back the car to Pedro), but obligatory in constructions where the dative argument carries the thematic role experiencer (e.g. *(Le) gusta la espinaca a Lucio ‘Lucio likes the spinach’), benefactor (e.g. *(Le) compré un juguete a Lucio ‘I bought a toy for Lucio’) or inalienable possession (e.g. *(Le) duele la panza a Lucio ‘Lucio has a tummy ache’) (cf. Demonte 1995, Cuervo 2003). Thus, the phenomenon of CLD represents a configurational mosaic within the Romance languages. Synchronic explanations of the distributional conditions that affect its emergence have been a great challenge until today (for a thorough discussion, see Anagnostopoulou 2005, forthcoming). The same holds for diachronic approaches. In recent years, several approaches have been presented (see Fontana 1993; Fischer 2002; Gabriel and Rinke 2010; Fischer and Rinke 2013; among others), each one explaining only parts of the development of CLD.

3. Clitic Doubling and the grammaticalization path of clitics One of the first and most influential approaches explaining the emergence of CLD has been based on the grammaticalization of the clitic pronoun. Fontana (1993) claims that the rise of CLD constructions in Old Spanish is connected to the grammaticalization process of the clitic: the loss of the complementary distribution (clitic or full DP but not both simultaneously) is related to the reanalysis of clitics from a Xmax to Xmin (1993: 224ff), i.e. from DPs to D-head categories. According to him, evidence for the categorial status of the clitic in Old Spanish can be found in interpolation configurations (3) (i.e. negation and also adverbs can be placed between the clitic and the finite verb) and in the distribution of the clitics (i.e. clitics appear in pre- and postverbal positions) (4):6 6

The data presented in this paper are part of the corpus CDAR which has been collected for the DFG research project “Clitic Doubling Across Romance” (FI 875/3-1) at the University of Hamburg (CDAR_Hamburg). It covers Spanish and Catalan texts from the 13th until the 19th century. For each century, two to four

120

(3)

(4)

Chapter Five

como a ty cierto es que As to you true is.3SG that lo non hamas. (OS) CL.3SG.ACC not loves “As you know that you do not love him.” CDAR_Hamburg [Corbacho_1438] e la emperadriu ha-li (OC) and the empress has.3SG-CL.3SG.DAT perdonat francament per ço cor Déus forgiven frankly because God e vós li avets feyta and You CL.3SG.DAT have.2PL done so tant d’onhor. much of-honour “And the empress has frankly forgiven him, since God and You have done him such honour.” CDAR_Hamburg [Crònica de Bernat Desclot_1299]

As to the status of Old Catalan clitics, the evidence is less clear cut. Fischer (2002) has shown that Old Catalan clitics never appeared left of negation (in her entire corpus no interpolation is attested), what is taken as a clear evidence for the clitic’s status as a D°. Nevertheless, they could appear preceding and following the finite verb in main and embedded sentences (5). This shows that the grammaticalization path of the clitics should rather be understood as a continuum; Old Catalan clitics, although less grammaticalized than the Old Spanish counterparts, should already be analyzed as heads. Furthermore, it should be pointed out that not all clitics in a specific period of a language are in the same stage, e.g. it has been assumed that accusative clitics in Modern Spanish (MS) are D-heads whereas dative clitics are already agreement markers (phi-features) (Bleam 1999).

different texts were chosen. The first 100 transitive sentences of each text were analysed and codified for several morphosyntactic features. The corpus has not been fully assembled yet. We would like to thank our student assistants Lisa Figura, Svenja Gottschick, Sarah Jobus and Clemens Kirsten for their invaluable work collecting and coding the data. See Corpus for the texts used.

The Clitic Doubling Cycle: A Diachronic Reconstruction

(5)

121

E diu que lo primer and said.3SG that the first respòs-li hòrreament e ab answered.3SG-him horribly and with males paraules. (OC) bad words “And he said that the first answered him horrified and with swear words…” (Fischer 2002; 13b/288)

The grammaticalization path of clitics can thus be represented as follows (6) (cf. Fischer and Rinke 2013: 467): (6)

Clitic DP

>

Clitic D-head

>

Clitic φ-features

As was argued by Fischer (2002) on the basis of Old Catalan data, it is not possible to find a univocal correspondence between CLD and the syntactic status of the clitic as a D-head. In a period where the clitic should be analysed as a D°, sentences are attested with and without CLD (7a-b). A satisfactory explanation of the emergence of such constructions requires therefore taking other factors into consideration, or even a combination of these. (7)

a.

b.

Prec-vos que m’ ojats tots a mi ask-you that me listen all to me un poc. (OC) a little “I ask you all to listen to me for a while.” (Fischer 2002: 43) e tan amarg és a mi and so bitter is to me que (...) (OC) that “And it is so bitter for me that (...).” (Fischer 2002: 44)

4. A new attempt We propose that the various patterns of CLD in the Romance languages are highly dependent on the interaction between the grammaticalization path of the clitic pronoun (from DP to D° to phi-

122

Chapter Five

features) (Fontana 1993, Fischer and Rinke 2013) and the syntactic behaviour of the full object DP, i.e. of word order changes. More specifically we claim that CLD represents an epiphenomenon emerging from the interaction between word order change and the grammaticalization process of the clitic and the full object DP. Fischer (2010), like many others before her (see Humboldt (1972 [1822]: 54f), Lehmann (1976: 455), and Givón (1979: 109)), has proposed that word order is related to information structure in OC like in the other Old Romance languages. This can be seen in the fact that full DPs appear in different positions triggering different interpretations, but also in the fact that the clitics can be placed pre- or postverbally, triggering a difference concerning information structure. Additionally, the relatively free constituent order has been subject to grammaticalization processes leading to the loss of discourse, pragmatic and semantic features formerly encoded through the linear order. As a consequence of this, the linear structure of the clause becomes more rigid, and the language system seeks other means to express the lost information. CLD would be a suitable mechanism to recover it. Under the view that grammaticalization processes involve cyclical changes (Gabelentz 1901 [1891], Jespersen 1966 [1917], Sapir 1921, van Gelderen 2004), we propose that the evolution of CLD in Romance languages can be represented by the following five stages, which can be reconstructed according to the data which are presented below: I. Absence of pronominal clitics (no CLD constructions). In this stage, only weak pronouns in the sense of Cardinaletti and Starke (1999) would be found, as a transition to real clitic pronouns. The position of the full object DP depends on the information structure of the clause, i.e. the object DP has quite some syntactic freedom: CLD is not found, even with personal pronouns, where it is obligatory nowadays. Old Romance texts with absolute absence of clitics are not attested. Since Latin lacks pronominal clitics, it is expected that there should be an early stage in Proto-Romance without this category clitic.

The Clitic Doubling Cycle: A Diachronic Reconstruction

123

II. Emergence of clitics (DP/D°). Restrictions on the placement of the full object DP start to appear. Since clitics are not completely grammaticalized yet, they still have a certain freedom concerning their placement w.r.t. the finite verb (4), and within the clitic cluster (8). The first optional CLD constructions appear in the language. Example (7) above shows a doubled full pronoun (7a) and the same pronoun without CLD (7b) (10): (8)

a.

b.

car lo pare lo because the father CL.3SG.ACC li havie tolt (OC) CL3SG.DAT had.3SG removed “Because his father had removed it to him” e el pages li and the peasant CL.3SG.DAT lo atorga CL.3SG.ACC concede.3SG “And the peasant granted it to him.” (Fischer 2002)

III. Grammaticalization of the clitic. The free-placement of the clitic w.r.t. the finite verb (see 4) and within the clitic cluster (see 8) is not allowed anymore. CLD of full pronominal DPs becomes obligatory. IV. Further grammaticalization of the clitic (semantic bleaching). The linear order of full object DPs gradually becomes fixed. CLD spreads to further contexts, e.g. animate nominal DOs.7 (9)

saludé a las maestras greeted.1SG to the teachers del jardín. (B.A. Spanish) from-the kindergarten “I greeted the teachers from the kindergarten.” (Zdrojewski and Sánchez 2014: 164) Las

CL3PL.ACC

7

Since full pronoun DOs in Spanish and Catalan can only refer to animate referents, CLD is linked to animacy in the first stages. So, the first nominal DOs which are doubled are also animate and only later on, CLD spreads to other kinds of objects, however, always according to the animacy hierarchy.

124

Chapter Five

V. Grammaticalization of the clitic continues (D°>phi). Case, person, number features are partially effaced from the clitic, which may than appear as the default form “lo”. The clitic being stripped from grammatical and information structural features allows CLD in all contexts (full pronominal DPs, animate and inanimate datives and accusatives with and without DOM). (10)

Eso también lo mata that too CL.3SG.MASC. ACC kill.3SG las plantas (Andean Spanish) the plant.FEM.PL “That too kills the plants.” (Caravedo 1996, apud Zdrojewski and Sánchez 2014: 165)

We assume that the next step would be the complete loss of the category clitic altogether, which is the starting situation of the whole cycle.

5. Corpus In this section, we will present data from Spanish and Catalan from the different five stages of the cycle showing how the emergence and spread of CLD can be correlated to changes in the status of the clitic itself and changes on word order and object placement.

5.1 Old Spanish/Old Catalan Already in the early stages, clitics are present in the language, fulfilling the same argumental function as full pronominal DPs (11a-b). The position of both, clitic objects and full DP objects, is connected to information structural requirements (e.g. Martins 1994, Fischer 2002). However, CLD constructions are still marginal (11c and 12). (11)

a.

pusieron a ellos put.3PL to them part (OS) side “They put them to one side”

a to

vna one

The Clitic Doubling Cycle: A Diachronic Reconstruction

125

b.

(12)

& enbio jª. mançeba que and sent.3SG a maid that la aduxiesse & CL.3SG.ACC bring.3SG and abriola. opened.3g-CL.3SG.ACC “and he sent a maid to bring it [the chest] and she opened it.” c. yo les fiz I CL3PL.DAT made.1SG saber a ellos to know to them “I let them know” CDAR_Hamburg [Fazienda de ultramar_1210] per mal che Mir Arnall for harm that M. A. m'avia fait et dict a mi CL.1SG.DAT-had.3SG done and said to me et a ma muler (OC) and to my wife “For the harm M.A. had done and said to me and to my wife.” CDAR_Hamburg [Grievances of Guitart Isarn_1080-1095]

From the 14th and 15th centuries, CLD appears more frequently, but it is still subject to specific contexts: full/strong personal pronouns (accusative and dative) (13-14) and quantifiers. (13)

a.

b.

otro que la amasse a ella other that CL3SG.ACC loves to her tanto commo él o so much as he or más (OS) more “other one who would love her like he, or even more” CDAR_Hamburg [El Conde Lucanor_1335] vós me fazedes a mí you CL.1SG.DAT make.2PL to me tanta merçed so much mercy “Sir, you make such a honor to me” CDAR_Hamburg [El Conde Lucanor_1335]

126

(14)

Chapter Five

a.

b.

que l'avia desafiat that CL.3SG.ACC-had.3SG challenged lo comte a ell (OC) the count to him “That the count has challenged him” CDAR_Hamburg [Epistolari d’Hipòlita_1549] de pagar-me a mi of paying-CL.1SG.DAT to me lo cque m'és degut what CL.1SG.DAT is dued “to pay me what is owed” CDAR_Hamburg [Epistolari d’Hipòlita_1524]

5.2 Standard Spanish/Standard Catalan Nowadays clitics have to appear before the finite verb (except for imperatives), postverbal clitics which have been possible in Old Romance (examples 4 and 5) are not attested anymore. CLD is obligatory with full/strong personal pronouns (accusative and dative pronouns) in both languages (15a and 16a). CLD with non-pronominal accusative DPs is still ungrammatical (15b and 16b). (15)

a.

b.

(16)

a.

Pedro *(lo) vio P. CL.3SG.ACC saw.3SG a él. (Stand. Sp.) to him “Pedro saw him.” *Lo vi al coche. CL.3SG.ACC saw.1SG to-the car “I saw the car.” Allà no *(el) vaig veure there not CL.3SG saw.1SG

The Clitic Doubling Cycle: A Diachronic Reconstruction

b.

127

a ell (només a ella). (Cat.)8 to him (only to her) “I didn’t see him there (only her).” Els zombies no paraven the zombies not stopped.3PL de mossegar(*-los) a tots els humans. to bite-CL.3PL to all the humans “The zombies didn't stop biting all humans”

However, dative arguments can be optionally doubled in some contexts (17), which can be taken as direct evidence for the ongoing grammaticalization of dative clitics, in contrast to the accusative clitics (see Demonte 1995, Cuervo 2003, Pineda 2013). In Spanish, number can be left morphologically unexpressed in some cases, indicating the beginning of semantic bleaching, i.e. the effacement of features. However, this is ruled out in Catalan, where dative clitics always show a different form for singular and plural. Furthermore, some Catalan speakers rate dative CLD as “rather unacceptable” (18).9 All this shows that the grammaticalization process of the Catalan dative clitics is less advanced compared to the Spanish clitics. (17)

(Le)

devolví gave.1SG back a Pedro (Stand. Sp.) to P. “I gave back the car to Pedro” CL.3SG.DAT

8

el the

coche car

The data for Modern Catalan are gathered using an Acceptability Judgment Task. The test consists of 26 items (11 sentences with CLD constructions under different conditions for case, quantification, definiteness, animacy, 10 distractors and 5 control items). The participants are asked to rate each sentence according to one of the four categories: “acceptable”, “rather acceptable”, “rather unacceptable”, “unacceptable”. They are also requested to correct the sentences which they judged to be unacceptable. So far, we have collected data from 327 speakers, mostly from the area of Barcelona. 9 Especially those speakers that have had the least contact with Spanish as a family language and in school.

128

(18)

Chapter Five

A la inauguración (li) at the inauguration CL.3sg.DAT van regalar flors a l’Ada Colau. (Cat.) presented.3PL flowers to A.C. “At the inauguration, they gave flowers to A.C.”

5.3 Spanish varieties In Buenos Aires (B.A.) CLD of accusatives seems to be less restricted than in Peninsular Spanish, i.e. doubling of accusative definite full DPs (19b), which usually show differential object marking (DOM) is also accepted (19a. In other Argentinean varieties, as in Patagonian Spanish, CLD of definite full DOs is also possible in the absence of DOM. According to Gabriel and Kireva (2012), these Spanish varieties have a more fixed word order, since almost 90% of the sentences in their corpus were SVO, and postverbal subjects were limited to unaccusative contexts. (19)

a.

b.

Ya los cumplí yet CL.3PL.ACC fulfilled.1SG a los once [años] (B.A. Sp.) to the eleven [years] “I am already eleven years old.” Lo agarré CL.3SG.ACC took.1SG (a)l mate. (Patagonian Sp.) (to) the mate “I took the mate.”

Lima Spanish even accepts doubling of accusative indefinite full DPs, usually with a specific reading (20). In Judeo-Spanish (JSP), we also find instances of CLD that do not correspond to the ones commonly accepted in Standard Spanish: in Judeo-Spanish CLD with animate definite full object DPs is optionally possible (21). (20)

Lo saludé a un CL.3SG.ACC greet.1SG to a estudiante que conozco. (Lima Sp.) student that know.1SG “I greeted a student that I know” (Zdrojewski and Sánchez 2014: 166)

The Clitic Doubling Cycle: A Diachronic Reconstruction

(21)

a.

b.

129

Yildiz disho ademas ke Y. said.3SG in addition that mi tiya Beki la my aunt B. CL.3SG.ACC ayudo a parir a su helped.3SG to bear to her madre. (JSP) mother “Y. said that my aunt B. helped her mother to give birth.” [eSefarad 01/09/2015] La hija hazina la the daughter sick CL.3SG.ACC yamó a la madre called.3SG to the mother “The sick daughter called her mother.” [eSefarad 20/11/2015]

Furthermore, Zdrojewski and Sánchez (2014) provide data from Andean Spanish showing that the morphosyntactic features of the accusative clitics have been bleached/neutralized (see also Navarro and Neuhaus 2016). (22)

a.

b.

Eso también lo that too CL.3SG.MASC. ACC mata las plantas. (Andean Sp.) kill.3SG the plant.FEM.PL “That too kills the plants.” Lo vendo toditos los CL.3SG.MASC.ACC sell.1SG all.DIM the carros. cars.MASC.PL “I sell all the cars.” (Caravedo 1996, apud Zdrojewski and Sánchez 2014: 165)

5.4 Typological application of the CLD cycle The data described in this section have shown that changes concerning CLD can be related to the other ongoing changes in Old Romance. Once clitics lose their “autonomy” (to appear pre- and postverbally), i.e. the possibility to express information structural features (Martins 1994, Fischer 2002), the first CLD constructions are attested. Once word order

130

Chapter Five

becomes fixed, CLD spreads to other contexts, according to the following hierarchy: full pronoun DPs > definite full DPs > indefinite full DPs. Summarizing: since grammaticalization implies cyclical changes, a clear path w.r.t. the evolution of CLD has been identified, a path which is also able to capture the distribution of different modern Romance languages along its hierarchy: I. Absence of clitics. The full object DP has some syntactic freedom depending on the information structure of the clause. 9 Latin, Proto-Romance. II. Emergence of clitics (DP/D°). The restrictions on the rather free placement of the full object DP increase. Clitics still have a certain freedom w.r.t. the finite verb, and within the clitic cluster. The first optional CLD constructions appear in the language. 9 Old Catalan; Old Spanish. III. Grammaticalization of the clitic, i.e. the free-placement w.r.t. the finite verb and within the clitic cluster is not allowed anymore. CLD of full pronominal DPs becomes obligatory. 9 Modern Standard Catalan; Peninsular Spanish. IV. Further grammaticalization of the clitic (semantic bleaching). Fixation of the linear order of full object DPs. CLD spreads to further contexts. 9 Buenos Aires Spanish; Lima Spanish; Judeo-Spanish. V. The grammaticalization of the clitic continues (D°>phi). In the clitics case, person, number, etc. get effaced. As the clitic is stripped from the grammatical and information structural features, it allows CLD in all contexts (full pronominal DPs, datives and accusatives with and without DOM). 9 Andean Spanish.

6. Summarizing and some conclusions In this paper, we have argued that the emergence of CLD constructions cannot be accounted for by focussing on a single factor, instead we have proposed that an approach that accounts for the interaction of grammaticalization and word order is necessary. We have shown that the status of the clitic, which has often been proposed as the decisive criterion to explain the emergence of CLD in the Romance language family, does

The Clitic Doubling Cycle: A Diachronic Reconstruction

131

not explain the distribution of CLD in the Romance languages. From the data presented in section 4, it has become obvious that CLD depends both on the grammaticalization path of the clitic (from DP to D° to phi-features) and on the changes concerning word order, i.e. the grammaticalization of the full object DP. Therefore, we claim that the emergence of CLD must be understood as an epiphenomenon of these other processes. Even more, CLD can be understood as a means of restoring information otherwise encoded in the linear order of the object w.r.t. the verb. Combining considerations on these two factors (grammaticalization of the clitic and the position of the object DP), it is possible to obtain a more fine-grained characterization of the CLD cycle, and to propose an explanation for the optional doubling, the doubling with personal pronouns and quantifiers, the extension to different types of full DPs, and the generalized CLD we observe in some languages. The proposed cycle can thus be applied to synchronic and diachronic data of Catalan, Spanish and their varieties. We believe that the diachronic application of the cycle to other Romance language is also feasible. However, this, as well as the explanation of the relation between the different grammaticalization processes within a theoretical syntactic model, remains for further research.

Corpus Texts used in the corpus CDAR_Hamburg Old Catalan: Llibre de meravelles, R. Llull (1288) Crònica, B. Desclot (1299) Epistolari de Sereneta de Tous (1374-1395) Lo somni, B. Metge (1399) La flor de les histories d’Orient, Aitó de Gorigos (1424) Les Corts de Jerusalem, anonymous (1424) Curial e Güelfa, anonymous (1432-1468) Sermons de Sant Vicent Ferrer (1449) Epistolari d’Hipòlita Rois de Liori i Estefania de Requesens (1524-1549) Memòries del cavaller valencià Gaspar Antist (1557) Old/Modern Spanish: Fazienda de Ultramar, anonymous (1210-1235) El Conde Lucanor, Don Juan Manuel (1335) Epistolario, E. de Villena (1396-1428) Corbacho, Arcipreste de Talavera (1435) Tratado en Defensa de las Virtuosas Mujeres, D. de Valera (1443) Generaciones y semblanzas, F. Pérez de Guzmán (1450-1455)

132

Chapter Five

La Celestina, F. de Rojas (1499) Epistolario, Conde de Tendilla (1504-1506) La Lozana Andaluza, F. Delicado (1528) Reprouacion de las supersticiones y hechizerias, P. Ciruelo (1530-1538) Lazarillo de Tormes, anonymous (1554) La infelice Dorotea, Lope de Vega (1622) Vida del capitán Alonso de Contreras, A. de Contreras (1630) El no importa de España, F. Santos (1667) Judeo-Spanish: eSefarad (01/09/2015), retrieved from on the 23th of October, 2015 eSefarad (20/11/2015), retrieved from on the 23th of November, 2015

References Anagnostopoulou, E., 2005, “Clitic Doubling”, in: M. Everaert, H. van Riemsdijk (eds), The Blackwell companion to syntax. Volume I, Oxford, Blackwell Publishing, 519–580. Anagnostopoulou, E., forth., “Clitic Doubling and object agreement”, in: S. Fischer, M. Navarro (eds), Proceedings of the VII Nereus International Workshop “Clitic Doubling and other issues of the syntax/semantic interface in Romance DPs”, Konstanz: Fachbereich Sprachwissenschaft Universität Konstanz. Aoun, J., 1981, The formal nature of anaphoric relations, PhD Dissertation, MIT. Belletti, A., 1999, “Italian/Romance Clitics: Structure and Derivation”, in: H. van Riemsdijk (ed.), Clitics in the Languages of Europe, Berlin, Mouton De Gruyter, 543–579. Bleam, T., 1999, Leísta Spanish and the syntax of Clitic Doubling, PhD Dissertation, University of Delaware. Caravedo, R., 1996, “Pronombres objeto en el español andino”, Homenaje al Dr. Germán de Granda, Anuario de lingüística hispánica, XII, 2, 545–568. Cardinaletti, A., E. Starke, 1999, “The typology of structural deficiency: a case study of the three grammatical classes of pronouns”, in: H. van Riemsdijk (ed.), Clitics in the languages of Europe, Berlin, Mouton de Gruyter, 145– 233. Cuervo, M.C., 2003, “Structural asymmetries but same word order: the dative alternation in Spanish”, in: A.M. DiSciullo (ed.), Asymmetry in Grammar. Vol. 1: Syntax and Semantics, Amsterdam, John Benjamins, 117–144. Demonte, V., 1995, “Dative alternation in Spanish”, Probus, 7, 1, 5–30. Dobrovie-Sorin, C., 1990, “Clitic Doubling, Wh-Movement, and Quantification in Romanian”, Linguistic Inquiry, 21, 351–397. Everett, D., 1987, “Pirahã clitic doubling”, Natural Language and Linguistic Theory, 5, 245–276.

The Clitic Doubling Cycle: A Diachronic Reconstruction

133

Fischer, S., 2002, The Catalan clitic system: a diachronic perspective on its syntax and phonology, Berlin, Mouton de Gruyter. Fischer, S., 2010, Word Order Change as a source of Grammaticalization, Amsterdam, Benjamin Publishing. Fischer, S., E. Rinke, 2013, “Explaining the variability of clitic doubling across Romance: a diachronic account”, Linguistische Berichte, 236, 455–472. Fontana, J.M., 1993, Phrase structure and the syntax of clitics in the history of Spanish. PhD Dissertation, University of Pennsylvania. von der Gabelentz, G., 1901 [1891], Die Sprachwissenschaft. Ihre Aufgabe, Methoden und bisherigen Ergebnisse, Leipzig, Tauchnitz. Gabriel, Ch., E. Kireva, 2012, “Intonation und Rhythmus im spanisch-italienischen Kontakt: Der Fall des Porteño-Spanischen”, in: E. Schafroth, M. Selig (eds), Testo e ritmi. Zum Rhythmus in der italienischen Sprache, Frankfurt, Lang, 131–149. Gabriel, Ch., E. Rinke, 2010, “Information packaging and the rise of clitic doubling in the history of Spanish”, in: G. Ferraresi, R. Lühr (eds), Diachronic studies on Information Structure. Language Acquisition and Change, Berlin, De Gruyter, 63–86. van Gelderen, E., 2004, Grammaticalization as Economy, Amsterdam, John Benjamins. Givón, T., 1979, “From discourse to syntax: Grammar as a processing strategy”, in: T. Givón (ed.), Syntax and Semantics, Vol. 1: Discourse and Syntax, New York, Academic Press, 81–112. von Humboldt, W., 1972 [1822], “Über das Entstehen der grammatischen Formen und ihren Einfluß auf die Ideenentwicklung”, Abhandlungen der Akademie der Wissenschaften zu Berlin, Berlin. Jaeggli, O., 1986, “Three issues in the theory of clitics: Case, doubled Nps, and extraction”, in: H. Borer (ed.), The syntax of pronominal clitics, New York, Academic Press, 15–42. Jespersen, O., 1966 [1917], Negation in English and Other Languages, Copenhagen, A.F. Høst. Kari, E., 2003, Clitics in Degema: a meeting point of phonology, morphology, and syntax, PhD Dissertation, University of Tokyo. Kayne, R.S., 1975, French syntax: The transformational cycle, Cambridge MA, MIT Press. Kayne, R.S., 2000, Parameters and universals, Oxford, Oxford University Press. Ledgeway, A., 2000, A comparative syntax of the dialects of southern Italy: a minimalist approach, Oxford, Blackwell. Lehmann, W.P., 1976, “From topic to subject in Indo-European”, in: N.L. Charles (ed.), Subject and Topic, New York, Academic Press, 445–457. Madeira, A.M., 1993, “Clitic second in European Portuguese”, Probus, 5, 155– 174. Martins, A.M., 1994, Cliticos na História do Português, PhD Dissertation, Universidade de Lisboa. Navarro, M., M. Neuhaus, 2016, “Clitic Doubling restrictions in Leísta Spanish”, in: S. Fischer & M. Navarro (eds.), Clitic Doubling and other issues of the

134

Chapter Five

syntax-semantic inferface in Romance DPs. Proceedings of the VII Neureus International Workshop. Arbeitspapier 128. Fachbereich Sprachwissenschaft, Universität Konstanz, 79–89. Pescarini, D., 2015, “The emergence of two classes of clitic clusters in (Italo)Romance”, Selected papers from LSRL 42, Amsterdam, John Benjamins. Pineda, A., 2013, “Double object constructions in Spanish (and Catalan) revisited”, in: S. Baauw, F. Drijkoningen, L. Meroni, M. Pinto (eds), Romance Languages and Linguistic Theory 2011: Selected papers from “Going Romance” Utrecht 2011, Amsterdam, John Benjamins, 193–216. Poletto, C., 2000, The higher functional field: evidence from northern Italian dialects, Oxford, Oxford University Press. Quícoli, C., 1976, “Conditions on clitic movement in Portuguese”, Linguistic Analysis, 2, 199–223. Rivas, A.M., 1977, A Theory of Clitics, PhD Dissertation, MIT. Sapir, E., 1921, Language, New York, Harcourt, Brace and company. Strozer, J.R., 1976, Clitics in Spanish, PhD Dissertation, UCLA. Torregrossa, J., 2012, Encoding topic, focus and contrast: informational notions at the interfaces, PhD Dissertation, University of Verona. Zdrojewski, P., L. Sánchez, 2014, “Variation in accusative clitic doubling across three Spanish dialects”, Lingua, 151, 162–176.

CHAPTER SIX A HISTORY OF PERSONAL SUBJECT PRONOUNS IN MILANESE IN COMPARISON WITH OTHER NORTHERN ITALIAN DIALECTS MASSIMO VAI Università degli Studi di Milano [email protected]

1. Introduction The Milanese dialect has undergone a great deal of change with regard to subject-pronoun organization: between the Middle Ages and the Modern era, it went through a period where the subject clitic (SCL) system was far more developed than it is now. I shall give consideration to the following issues: (i) like other medieval varieties, the Milanese dialect went through a period of asymmetrical pro-drop, whereby the pronominal subject may be lacking in main clauses, but it is almost always present in embedded clauses; clitic complements follow Tobler-Mussafia (TM) Law; (ii) during the medieval stage of Milanese dialect, stressed complement pronouns, which would become modern subject pronouns, were also used as subjects on the left periphery of the sentence; (iii) in the documents dating from the 14th-15th centuries a process begins whereby the order between subject pronouns and pre-verbal negation changes; the first evidence of clitic a also appears; (iv) in the same period, the reduction of instances of Tobler-Mussafia (TM) enclisis detracts from speakers evidence for Verb Second order: this evidence is lacking on account of the occurrence of a clause structured with the order: “sì - complement clitic - inflected Verb” and the reduction of apparent XVS word order: the only evidence of V2 occurs in presence of some initial particles (e.g. mò, donca, etc…);

136

Chapter Six

(v) in the 17th century, the Milanese dialect has a SCL system much richer than now: the presence of SCLs in the 1st, 2nd, 3rd and the 6th person (as far as the 4th is concerned, the SCL of the 1st person applies); by the end of the 18th century, SCLs system has acquired a typology similar to that of contemporary Milanese except for interrogative VS order which still remained (and now is gone).

2. 2nd half of 13th century: Bonvesin dra Riva’s Vulgaria At this stage, as in other Medieval Romance languages, the basic word order is SVO, which can be found in dependent clauses; in main clauses a derived order XV (…) of verb second type (see Benincà 2004: 261 and references) occurs: (1)

H 213

Bon vin fa l’ uva good wine makes the grape “Black grape makes good wine”

negra black

(2)

A 205

per lu sont eo For him am.1SG I “I am a queen for him”

(3)

T67

E anc de mi – diz and also of me – says quello – sempre á Zené beffao that-one – always has January mocked “And of me too—he says—January has always mocked”

regina queen

In (1) the object (bon vin) is immediately followed by the verb without resumptive clitic due to the V2 syntax: this is a feature shared by all medieval Romance languages. Pragmatically, it can have various interpretations: contrastive or unmarked focus, but also theme: see Benincà 2004: 267. In (3) the order á Zené beffao is noteworthy: modern Romance languages allow the order AUX-SUBJ only with clitic subjects, with few exceptions.1 In this case Old Milanese behave like Old French.2

1

Benincà (1994: 126) cites an example from Ladin of Val Badia: Plö tert à Col fat pert dla comunité de La Pli (lit.: “later has Col made part of etc”). Here Col is a nominal subject 2 Benincà (1994: 127) cites: Un peu après eure de prime fu Mador venuz a cort.

A History of Personal Subject Pronouns in Milanese

137

2.1 Asymmetrical pro-drop In Medieval romance languages, V2 creates a type of asymmetrical pro-drop (see Benincà 2004: 263 and references): the subject may not be phonologically realized in main clauses, while it is generally realized in dependent ones: (4)

Q65-66 Quand tu veniss al mondo, when you came.2SG to-the world se tu voliss pensar if you wanted.2SG think-about.INF negota ge portassi pro, nothing there brought.2SG negota n poi pro portar nothing from-there can.2SG bring-INF “When you were born, if you would think on it, nothing you brought with you, nothing you can bring away”

In medieval Northern Italian dialects, the XP (or more XPs) which precedes the verbs in main clauses may be located in the various projections of a split-CP situated in the left periphery (Benincà 2004: 2701 quotes examples also from other medieval Romance languages): (5)

[per tug li tempi] me rend pro [to her] [for all the times] me surrender.1SG e me consegno. and me give.1SG “Now and forever I surrender and give myself to her”

S III 372 [A lé]

(5) shows that the V2 syntax of Italian Romance appears less rigid than in other Romance languages: V1, V3, V4 is very common in all the languages of medieval Italy (see Benincà 2004: 275); [A lé] and [per tug li tempi] are constituents located in the left periphery of the sentence.

2.2 Position of clitic complements: Tobler-Mussafia Law As in other medieval Romance languages, clitic complements follow Tobler-Mussafia Law, so, for example, there is no fixed position for clitics that combine with the imperative mood:3 3

See also Dante, Purg. I, 81-82: “per lo suo amore adunque a noi ti piega / Lasciane andar per li tuoi sette regni”.

138

(6)

Chapter Six

I104-105

Donca argorda ’t del so remind.2IMP you of-the ben tu he goodness you have.2SG per mi trovao / by me found Apress zo te dementega after that you forget.2IMP del mal k’ è of-the evil that is strapassao gone “So remember the good you found through me/ then forget the evil that is gone”

According to Benincà (1994: 232), two conditions should be observed in order to have enclisis of the complement pronoun: (i) it is necessary that the verb be shifted to C° and (ii) if the SpecCP (SpecFocP in cartographic terms, see Benincà 2004: 275) is empty, there is enclisis. On the contrary, when SpecFocP is occupied by phonologically realized or abstract elements, enclisis is impossible, as shown in (7)-(8) (from Monumenti del dialetto di Lio Mazor): (7)

8.t.284

(8)

17.r. 175

me uoj-tu dar to-me want-you give.INF la tauerna? the tavern “Do you want to give me the tavern?” et così lo mis-e’ ço and so it.OBJ put-I down “And so I put it down”

In the first example, SpecCP should be occupied by a wh-operator, in the second one SpecCP is occupied by a “filler of SpecCP”, which is often the particle (co)sì. Benincà (2004: 285) reformulates the TM generalization: “When the verb is in C° and is preceded by an element that has to be in a Spec of the Focus field, enclisis is impossible”.

4 5

Levi (1904: 23). Levi (1904: 31).

A History of Personal Subject Pronouns in Milanese

139

2.3 Reciprocal order of complement clitics and subject pronouns When complement clitics and a subject pronoun co-occur in the same main clause, they occupy different positions: (i) in most cases the subject pronoun in the nominative case is placed immediately before the complex: (NEG)—complement clitic—inflected verb: (9)

A 321

Da po ke De saveva since then that God knew.3SG anz k’el m’ havess creao before that he me had.3SG created “As God knew before he created me”

(ii) when the subject pronoun immediately follows the verb on account of Verb Second syntax, the clitic complement generally occurs between the initial XP (and after the negation particle, if there is one) and the inflected verb: (10)

A 48

(11)

L 120

Per quel no ʼt for that not to-you eo torto I wrong “So I don’t do wrong to you” Nïent ghe vol el nothing to-him wants he “He wants to do nothing for him”

faz do.1SG

far do.INF

As stated above, complement clitics must be placed according to TM Law, which is probably the Old Romance result of Wackernagel’s law (Wackernagel 1892)—see Salvi (2004: 24)—i.e. the Indo-European law which rules processes of enclisis. On the contrary, the occurrence of subject clitics in the Romance varieties in which they appear seems to be related to another process (as already hypothesised by Meyer-Lübke 1897: 333): pre-verbal subject pronouns, formerly in the nominative case, gradually “draw close” to the structure of verbal inflection; more specifically, Northern Italian dialects seem to have reanalyzed subject clitics as a “competing head that replaces some of Agreement syntactic functions” (see Poletto 1995: 296). As for post-verbal subject pronouns, which will lead to the formation of an “interrogative conjugation”, their

140

Chapter Six

position is a consequence of verb movement due to the V2 syntax of interrogative clauses. If this hypothesis is correct, the process that led to the development of clitic pronouns having subject function has nothing to do with complement clitics subject to TM Law: Romance subject clitics were probably never ruled by TM Law (see also Colombo 2016: 166).

2.4 Oblique personal pronouns used as subject in Bonvesin Oblique pronominal forms could also be used as subjects in particular contexts. These contexts have in common the fact that the pronominal subject probably is not in SpecAgrP: (i) with clausal gerunds,6 e.g.: (12)

P251

Sapiand lu ke ʼl by-knowing him that the demonio zo feva a tal devil that did.3SG in such tenor way “As he knew that the devil did so in that way”

(ii) in case of subject in the left periphery,7 e.g.: (13)

B 504

Conven ke lu dai It-is-good that him by medici devess fi medegao doctors should be treated “It is good for him to be treated by doctors”

More precisely, the form lu (lit. “him”) of non-nominative origin, appears in about forty examples as a subject:

6

Pires (2006: 58-59) notes that subjects in absolute clausal gerunds could receive either Nominative or Accusative case in English, so the determination of specific case morphology of subjects in clausal gerunds seems to involve more complex mechanisms than a single default case, e.g.: Mike expected to win the game, he/him being the best athlete in the school. 7 Pires (2006: 59) notes that a similar possibility also exists for certain topic DPs in English, which are not realized in a standard structural case position, e.g.: Mike/him, I have never met.

A History of Personal Subject Pronouns in Milanese

(14)

T 20

141

Lu malfazando regna him by-wrong-doing reign.3SG “He reigns by wrong-doing”

Bonvesin’s language allows null subjects, so sentences in which lu behaves like a subject can be consistent with the asymmetric pro-drop displayed by medieval languages: we can assume that lu is placed in a position different from SpecAgrP, as shown by the fact that lu may appear before el (see below), while pro should be in SpecAgr: (15)

P 1-4

Eufimïan da Roma Eufimian from Rome fu nobel cavalé Poënt era-pro was noble knight Mighty was.3SG e richissimo e molt and very-rich and very amig de De. Ai peregrin, friend of God to-the pilgrims ai poveri molt era-pro lemosné: to-the poor very was.3SG almoner Per questa via teniva-pro e In this path kept.3SG and lu e la muié him and the wife “Eufimian from Rome was a noble knight, he was mighty and very rich and very dear to God. He was very charitable to pilgrims and poor people; both he and his wife used to keep to this path”

Similarly to lu, the oblique form lor can also be used as a subject: (16)

P 92

E and se REFL

lor in them to ʼn therefrom

divers parte different parts van-pro per go3PL to

lu him

trovar find.INF “And they go in different directions to find him” However, el is still used as a subject as well: most of the instances of a 3rd person subject in Bonvesin appear in the following context:

142

Chapter Six

(i) (ke) - el—(NEG)—complement clitics—inflected verb: (17)

T 50

E’ squasso giaza e neve I shake1sg ice and snow k’ el m’ á lasá per that he to-me has left for pegno pledge “I shake the ice and snow that he left me as a pledge”

(ii) inflected verb—el/’l (V2 or interrogative context): (18)

N 76

Anc n’ abia el ben even-if of-that have.SUBJ3SG he well d’avanzo,perzò no dé ʼl enough,for-that not must he fá stragio make. INF massacre “Even if he has more than enough, notwithstanding he must not make a massacre”

(iii) el occurs in the following context in fewer than twenty sentences too: el—XP* - inflected verb: (19)

I 29

La passïon k’ av Criste the Passion that had.3SG Christ e k’ el per ti and that he for you portava bore.3SG “The Passion which Christ had and bore for you”

(19) shows that there can be a constituent between el and the inflected verb: so we can think that el is not yet a clitic pronoun in Bonvesin’s language.

2.5 Co-occurring of lu and el in the same sentence Taking into account that ancient nominative pronouns would become subject clitics in modern dialects, at least two examples in Bonvesin’s text are very interesting:

A History of Personal Subject Pronouns in Milanese

(20)

T 80

(21)

P 179

143

Lui malfazand, eli ten him by-wrong-doing he keep.3SG nu oltri in servitura us others in servitude “He keeps us in servitude through wrong-doing” Lui ʼn voiand vana gloria him not willing vain glory eli è da illό partio he is from there left “He, unwilling to vainglory, left from there”

In these contexts lu and el are co-referent within the same sentence: lu, which is within a dependent clause in the gerund mood, is likely to be placed in Topic Field and is co-referential with the pronoun el, which can be in SpecAgrP of the main clause. This context looks like an antecedent stage of a case of Vanelli’s (1988: 55) “reduplication”, that is the co-occurence of a free subject pronoun and the corresponding subject clitic pronoun (ti te parli and sim.) of modern varieties. In order to explain the difference between the old stage and the new one (see Vanelli 1998: 64-65), a process of re-analysis of this context—where lu should be thought formerly in a TopP and el in SpecAgrP—is likely to be necessary as claimed in Roberts (2007: 39-40). This stage led in turn to a new one in which lu is in SpecAgrP and el in a more complex Agr° (Poletto 1995: 314). The now observable variation between Northern Italian pro-drop and non-pro-drop dialects should depend on the analysis of lu as placed in a SpecTopP or in SpecAgrP: (22)

OLD SYSTEM

MODERN SYSTEM

TopP TopP 3 3 CP AgrP Ÿ CP AgrP 6 3 6 3 lui (malfazand) SpecAgr lui (malfazand) SpecAgr Agr' | | 2 eli (lui) eli

144

Chapter Six

Table 1: Bonvesin’s subject pronoun system according to Contini’s 1941 edition

1 2 3 4 5 6

Stressed obl. pronouns used as subjects mi lu / le

lor

Nominative pr.— XP—V

Nominative pr. in SpecAgrP

eo tu el / ella nu vu i

eo, e’ tu, ʼt el, ʼl, ’lo / ella, ’la nu vu, voi i /el

2.6 Bonvesin’s language in Contini’s 1941 edition compared to other editions So far we have taken into consideration Contini’s 1941 edition of Bonvesin’s Vulgaria. However, Bonvesin’s language, as edited by taking into account other variants from other manuscripts, presents various kinds of subject pronouns, which already show signs of an evolution that will lead to the subsequent system. In Wilhelm’s 2006 edition of St. Alexis— based on Trivulzianus 93—(here W), Wilhelm notes that Contini’s edition of the same text (here C) contains an inventory of pronominal subjects (or in subject function) that is very small compared to the multiplicity of variants documented in the manuscript from Milan, e.g. for 3rd persons: (23)

(24)

W 198 Eufimian so padre al ave C 198 Eufimïan so patre el av [Eufimian so padre].OBJ he had “he had met (there) his father Eufimian” W 94 che quele terre onde that (in) those lands where ay van facen pregonamento SCL.3PL go.3PL make.3PL call C 94 k’in quel terr o van “That they should make a proclamation where they go through”

incontrado illó incontrao (there) met

in those lands

A History of Personal Subject Pronouns in Milanese

145

Notice that some forms that are apparently weakened (al, ela, ay) appear in SpecAgrP, which seems to anticipate further developments, i.e. the development of a system of subject clitics. Table 2: Personal subject pronoun system in Bonvesin’s St. Alexis (Wilhelm’s edition)

3

Stressed oblique pronouns used as subjects lu, luy / le

Nominative pr.— XP—V el / ella, ela

6

lor, loro

i, ei

Nominative pr. in SpecAgrP el, al, i, ʼl, ’lo / ella, ela ’la i, illi, il, eli, ei, ay, ai/el

3. Documents between the 14th and 15th century Beginning with the 14th century, the language policy of the Visconti Court has been oriented towards the Tuscan dialect, so for the period between the 14th and the 15th century we can find mostly few texts on a religious topic, both in poetry and in prose, which are written in a Western Lombard dialect without any other linguistic specification of their origin, except for one, which Colombo (2016: 17) believes should be classified as Old Milanese. They are Margarita Lombarda, (Meditazione sulla) Passione, Esposizione del decalogo and Passione trivulziana. Margarita Lombarda is a poem dating from the end of the 14th century. In this text main clauses are still analyzable in terms of V2, which is however often difficult to recognize, because: (i) there are few cases of apparent V2 with phonologically realized subject; (ii) the instances of Tobler-Mussafia enclisis, which should be possible only if the inflected verb is already in C° (see Benincà 1994: 232; 2004: 275) and therefore should constitute positive evidence of V2, are rarely present because of the frequent use of lexical elements like e.g. sì/se < SIC, which trigger proclisis of clitics, instead of enclisis (see Benincà 2004: 266). Examples of TM enclisis:

146

Chapter Six

(25)

325

(26)

638

e va se ʼn and he-goes (him)self from-there denanze al so dé before to-the his god “And he goes before his god” E fazo ge fare and make.1SG to-them make.INF molti peccady many sins “And I make them commit many sins”

Sentences introduced by (XP)(e) sì (sè) < (ET) SIC, which triggers proclisis: (27)

21

Sì l’ adorava como so it.OBJ worshipped.3SG as dé god “So he worshipped it as his god”

so his

(iii) in preverbal position, arguments different from the subject are rarely attested: (28)

345

(29)

649

E cossi po tu anche and so can you also guarire heal.INF “And so you can also heal” perzo m’ e therefore me have.2SG fato cossi tristi made so sad “Therefore you have made me so sad”

tu you

3.1 Subject pronoun—NEG—V > NEG—subject pronoun—V At this stage, the negative particle is still placed between the subjectpronoun and the inflected verb:

A History of Personal Subject Pronouns in Milanese

(30)

657

147

Ch’ e’ no t’ that I not you olza avri la bocha hear.1SG open.INF the mouth “That I should not hear you opening your mouth”

However, there are also some instances where we can find the order NEG - subject pronoun—verb: (31)

599

si che no l’ so that not she miga pagura NPI8 fear “So that she had no fear at all”

ave had.3SG

From the point of view of geolinguistic considerations, the change from the order subject pronoun - NEG—verb to the order NEG—subject pronoun—inflected verb occurred in a continuous dialectal area which includes Lombard dialects as well as Venetian ones, but it was never encountered anywhere else, i.e. it never included the Piedmontese dialect or Emilian dialects. In Margarita we first find an incoherent order subject pronoun-NEG / NEG-subject pronoun. in case of 3rd person singular forms, thereafter the NEG-pronoun-verb order extends consistently in the 3rd person singular but not yet in 2nd singular, while towards the end of the 17th century the change is completed in Milanese for all the verballyinflected persons. With other XPs, the position remains unchanged (in particular with the forms of the stressed pronoun), so this change seems to be a clue of the structural greater “proximity” of pronominal forms to the verbal inflection, and then a clue of their clitic status, see Rizzi (1986: 398): “[…] the negative clitic and the subject clitic are members of the same clitic cluster”.

8

Negative polarity item.

148

Chapter Six

Table 3: Inventory of pronominal forms used as subject in Margarita Stressed oblique pronouns used as subjects 1 2 3

luy

4 5 6

loro

Preverbal subject pronouns

Postverbal subject pronouns

eyo, ei’, e’ tu el, al, l’, i’, e’ / ella, ela, ala, la, el’, al’, l’ nu vuy, vu eli, il, ai, li, (e’?) / le

é-tu, ve-te, ví-to fè-l

aví-vo

3.2 Appearance of the clitic a In Margarita this verse is pointed out by the editors (Wilhelm et al. 2011: 148): (32)

117

po’ ch’ a’ no as that a NEG de so piazimento of her liking “As it isn’t to her liking”

l’ it

è is

In Vai (1996: 70) I pointed out that the presence of the preverbal negation particle in the Prissian allows the clitic a component to be isolated within the clitics of 3rd person singular and 3rd person plural, so: al/ai beside a-no-l / a-no-i, so that we can hypothesize that these pronouns are made up of two parts: an invariable clitic particle a plus the personal subject clitic components l/i, cf. Benincà (1994: 121-2).9 In this way Margarita makes it possible to backdate the evidence of this analysis to two centuries earlier. Colombo (2016: 163-167) cites other instances of al, ai (and a no l, a no i) from old Northern Italian dialects. There are two major hypotheses about the origin of the clitic particle a: 1st hypothesis: origin from a pronoun Lomb. a = Florentine e < ILLE or similarly.10 From a syntactic point of view, the comparison with the 9

According to Salvioni ala = a + la: “ala dis” is not different from a te diset […]; a goes together with te, but it doesn’t replace it”. 10 We can think to the following path – el > e > a or el > al > a for Milanese, which presupposes phonetic erosion due to the high degree of grammaticalization of this

A History of Personal Subject Pronouns in Milanese

149

Florentine particle e seems to be plausible, particularly with the “analogical” e—distinct from “primary” e—according to Brandi e Cordin (1981: 75), since this e may co-occur with subject clitics too, e.g.:11 (33)

e 'dɔrmo (e) tu d'dɔrmi e 'dɔrme / (e) la 'dɔrme e si 'dɔrme (e) vu ddor'mihe e 'dɔrmano / (e-l)le 'dɔrmano

e ɔ ddor'miho (e) t a ddor'miho (e)ʎʎ/la ddor'miho e s ɛ ddor'miho (e)v a'vehe dor'miho ʎʎ/lanno dor'miho

There are however some difficulties in considering pronoun el as antecedent to clitic a: while in Milanese the particle a seems to be a standalone element already by the late 15th century, there is little evidence of a phase where e is similarly extended to all the pronominal persons: in actual fact, in Milanese e is the clitic stable for the 1st person singular, while we can observe few definite instances of 3rd person singular e < ILLE/ILLI: an uncertain instance in Fabio Varese II.1.9 -10: Par mezz ai beccarij, par mezz al foss / e se sent i becché co’ i sû folsciasg “between butchers and ditches, one can hear butchers with their knives”; another instance in Prissian 163.1: i Latin antighament e fauenn ben a scriu i longh “formerly Latins were right to write long vowels”. There are indeed some instances in other Lombard dialects and only one certain example in Margarita 669 12 “where he (e) made the great war”. 2nd hypothesis: Zerreissung of al. Lorck (1893: 164) thought that in the old dialect of Bergamo, where he found cases similar to those we have seen so far in Milanese, a subdivision (Zerreisung) should have occurred in the 3rd person singular and plural pronouns, so that the subject pronoun al/ai was divided in two constituents and the negation particle no occurred in between: (34)

III 157 Per que a-no-la-y volse consentire13 = ala no ye “Because she didn’t (a-no-la-y) want to allow them”

pronominal form. Colombo (2016: 161–162, n. 100) hypothesizes some kind of phonological reduction on the basis of cases like “e tanto com più la persona ama la cossa quando e’ la posede, de tanto el è più dolore quando el la perde”, maybe through reanalysis. 11 From Manzini and Savoia (2005). 12 Wilhelm et al. (2011: 69). 13 Lorck (1893: 73).

150

(35)

Chapter Six

V 62

Per che a-no-y vols in lu credi14 = ay no “Because they did not (a-no-y) want to believe in him”

3.3 Meditazione sulla Passione and Esposizione del Decalogo (from Como) A Meditation on the Passion written in prose, included in a 15thcentury codex discovered in a library in Como is also noteworthy in our analysis, because in that period written texts from Como were linguistically quite similar to Milanese15 (indeed, linguistically speaking, this text only contains West-Lombard linguistic features). Also an Exposition of Decalogue was found in the same codex. In these texts some cases of V2 may still be observed too:16 (36)

4.22

(37)

6.6

Doncha era-lo inimigo no therefore was.SCL.3SG enemy not amigo friend “Therefore he was an enemy, not a friend” Mo i uo fagio bene now have.2PL SCL.2PL done well “Now you have done well”

So, in these and other instances V2 is determined by the occurrence of certain elements: doncha, mo, ben, or, qui, illora.

14

Lorck (1893: 78). See Stella (1994: 194-195). 16 Numbers refer to the pages and lines of AGI IX. 15

A History of Personal Subject Pronouns in Milanese

151

Table 4: Inventory of pronominal forms used as subject in Passione (and Decalogo) Stressed oblique pronouns used as subjects

Preverbal subject pronouns

Postverbal subject pronouns

1 2

mi ti

eio, ei’, e tu

3

lu

elo, el, al, ell-, -ll- , e

o-e e-tu, si-tu, po-tu, distu, poris-tu a-l, e-llo, era-lo, fosselo, respox ello / domanda-la, uiuera-la

4 5 6

loro

nui uu illi, li, el, ei?

i-uo in-li

3.4 Passione Trivulziana The text of another Passion is part of a Codex from Biblioteca Trivulziana of Milan (Codex Trivulziano 1993). Colombo (2016) believes that this text is to be dated in a period from around the 14th century to the first half of 15th century, and the place of composition (on the basis of phonological and morphological features) should be Milan or the surrounding area. Also in this text we can see some cases of residual V2 after ben, doncha, mo, quilò (but less systematically than in Passion from Como):17 (38)

87 109 121 140

Mo poris tu dire “Now you could say” (lit.:now could you say) e ben lo poì-vo vedere “And you can indeed see it” (lit. and well it can-you see) e s’el vore morire doncha è-’l mato “And if he wants to die, so he is mad” (lit.: so is he mad) Quilò poris tu dire “Here you could say” (lit.: here could you say)

In this text we still find examples of TM enclisis:

17

Numbers refer according to Colombo (2016).

152

(39)

Chapter Six

7

133 171

Guardòsse l’un l’altro “They looked at each other” (lit.: looked.3SGOBJ.CL.REFL) e nudrigòve “And he fed you” (lit.: fed-2SG.OBJ.CL) e pià una sponga e bagniàla “and he took a spunge and wet it” (lit.: and wet3SG.OBJ.CL)

Like in other medieval texts, the context for TM enclisis is often removed because of the presence of particle sì, which triggers proclisis, as in: (40)

131

Toiìlo vu e sì lo crucifiché. “Take him yourselves and crucify him” (lit.: him-crucify)

Here we have enclisis with IMP.2PL Toiìlo, but proclisis with IMP.2PL crucifiché, due to the presence of sì. Taking into account that main clauses are often introduced by sì after a topicalised constituent (also consisting of a clause, e.g. as is the case for clauses introduced by quando “when”), the reduction of instances of Tobler-Mussafia (TM) enclisis, due to the presence of the particle sì, detracts from speakers positive evidence of V2 order (the resulting order being sì - complement clitic - inflected verb, with proclisis). Colombo (2016: 163) analyses a’’l / a’l’ / a’y as composed of a+l, a+i too, and interestingly notices the minimal pair: (132) “s’ela no te fosse dada” vs. (153) “ke no la podesse cadere”, where we can see the order ela no against no la in the same text, i.e. the same change of order synchronically that we will observe diachronically in the history of Milanese.

A History of Personal Subject Pronouns in Milanese

153

Table 5: Inventory of pronominal forms used as subject in Passione Trivulziana

1

Stressed oblique pronouns used as subjects mi

2

ti

3

lu18

Preverbal subject pronouns

e-to, vo’-tu, vé-tu, sé-tu a-l, era-lo, è-l, disse-lo

elo, ello, ell’, e19 nu vu, vui, vuy, (vu altri)

loro

Postverbal subject pronouns

e’, eiio, ei’, ey’, i’, io tu a

4 5

6

a

savì-vo, determinévo, avise-vo, sì-vo, vorì-vo, poì-vo

illi, elli

4. Lancino Curti (1460-1512) As said before, the language policy adopted by the Milanese Court inclined to the prominence of Tuscan as a language, so that written Milanese started to be used only on parodistic and playful occasions. At this stage, in Lancino Curti’s Sonnets clitic subject pronouns e/i and te of 1st and 2nd singular also appeared, which were morphologically differentiated from tonic mì and tì:20 (41)

III.8

Te

vedaré t’ è will-see.2SG SCL.2SG have.2SG habiù un bel cermeson had.PA.PPLE a fine head “You will see that you had a large head” SCL.2SG

18

(62): “ke per quelo lu era venudo al mondo, azò k’el rezevesse morte”. (163): la persona ama la cossa quando e’ la posede. 20 Lancino Curti’s texts are quoted (with same numbering) according to Isella (1979). 19

154

Chapter Six

(42)

IV.6

(43)

I.7

(44)

I.15

Avè ardiment de vorè have.INF impudence of want.INF tì dì mà you speak.INF ill “Having the impudence for you to want to speak ill” e so ch’ avẹreve SCL.1SG know.1SG that have.COND.1SG an mì quai cos sgià also I something already scrig written “I know that I already would have written something too” Quel ch ’i ò what that SCL.1SG have.1SG scrig i ò qui written SCL.1SG have.1SG here “I have here what I have written”

Sometimes clitic a spreads, also substituting other clitics: (45)

II.9

(46)

IV.5

Dison ch’ a in sempiedẹ say.3PL that a are.3PL craps tut quel ch’ al dix 21 all that that SCL.3SG says “They say that all he says are crap” A ʼt par a ti, a to-you.CL seems to you, sbirascịọ, un bel mesté? servant(?) a good thing? “Does it seem to you a good thing, menial servant?”

5. The Rabisch (1589) and compà Baciòcch’s frottole The Rabisch (“Arabesques”) of the “Milanese Academy of Blenio Valley” (headed by Giovan Paolo Lomazzo) is a collection of poems written by the associates in ‘facchinesca’ language (a kind of artificial dialect similar to that of the porters who came to Milan from the Ticino valleys, in particular from the Blenio Valley). Essays from 61 to 64, by the 21

Otherwise: a + SCL.3SG.

A History of Personal Subject Pronouns in Milanese

155

academician Jerome Maderno (academic nickname: Compà Baciòcch), are written in a Milanese folk dialect. The inventory of subject clitics found in Maderno is substantially similar to the forms that are attested up to Maggi’s works: the persistence of the 1st pers. sing. clitic e/i (sometimes replaced by the clitic a), 3rd pers. pl. clitic i, frequent use of the clitic a even together with personal pronoun clitics:22 (47)

II, 61 2-3 i’

o sentù on gran have.1SG heard a great spavent / Dov’ e’ cred che fright / where SCL.1SG think that malcontent discontent “I have heard a big fear, where I think that discontent”

SCL.1SG

(48)

II, 64 25 E

(49)

II, 61 31-32

22

se a i and if a SCL.1SG ham da fa on have.1PL to do.INF a lavó / work El farem s’ OBJ.CL do.FUT1PL if el poram fa OBJ.CL can.FUT1PL do.INF “And if we have to do a work / we ‘ll do it if we can” S’ a i ghe if a SCL.3PL IND.OBJ.CL dan per sòrt on scròl / give.3PL by chance a shake A i ghe a SCL.3PL IND.OBJ.CL vûn mett su dra sa want.3PL put.INF on of-the salt “If, by chance, they give them a shake, they want to put salt over them (i.e.: “they will preserve them in salt”)”

Numbering according to Isella (1993).

156

Chapter Six

5.1 Contexts which seem to promote the occurrence of clitic a These contexts seem to promote the occurrence of the clitic a: (i) main clause preceded by dependent clause: (50)

II, 61 325

s’ o fallà if have.1SG erred me ne pent me of-it.CL regret.1SG “If I made a mistake, I regret it”

a a

(ii) (hanging) topic: (51)

II, 61 230

I nodé del criminal / the notaries of-the criminal (trial) Ai ghe a SCL.3PL IND.OBJ.CL vûn taià le al want.3PL cut.INF the wings “As for the notaries of criminal trial / they want to cut their wings”

Sometimes a occurs in contexts more similar to that which was observed by Benincà for Paduan a, that is at the beginning of a brand new sentence: (52)

II, 63 88 sgg.: PEDRETT A te a IND.OBJ.CL2SG po’ el nas po23 the nose “I will break thy nose” ZAN A te a IND.OBJ.CL2SG el gavasc the throat “I will break your throat”

23

romparò break.FUT1SG

romparò break.FUT1SG

Perhaps a particle like those analyzed in Poletto and Zanuttini (2010)?

A History of Personal Subject Pronouns in Milanese

157

5.2 Placement of SCL co-occurring with the negative particle As it has already been pointed out, the order inside the complex us some clues about the structure of AgrP:

NEG—SCL gives

(53)

II, 61 96

(54)

II, 61 183

A no i vûl a NEG SCL.3PL want.3SG che ona baretta / that one bar “They don’t want one bar to be…” i ho pû intes, SCL.1SG have.1SG also heard che no i è foll which NEG SCL.3PL is fibs “I have also heard, and they are not fibs…”

In about the same period, we can get some information by means of a translation of Boccaccio’s tale “The King of Ciprus”, as having been translated into a few Italian dialects by Lionardo Salviati in his Avvertimenti della lingua sopra’l Decameron, published in 1584: (55)

In lingua Milanese: […] che nol’auerau fac negotta: perche ol Re era tant dapuoch, che nol feua gnanc ment a inghiuri che gheren’ fag a lui In lingua Padouana: […] era d’una uita si sdramazza, e così da puoco ben, che ello no solamentre el no fasea uendetta […]

Here we have further confirmation that at the end of the 16th century the process SCL-NEG > NEG-SCL (at least, in the case of the 3rd person singular) was already completed in Milanese, while Paduan still seems to not display it.

5.3 3rd person plural verbal forms distinguished from 3rd person singular forms only by subjectclitics In Maderno there are numerous instances of 3rd pers. pl. verb forms identical to the 3rd pers. sing. As for the modern Milanese dialect, this is possible only when i) an unaccusative verb precedes the subject or ii) in subject relative clauses. In Maderno, other contexts allow this kind of

158

Chapter Six

agreement too, provided that the 3rd pers. sing. inflected form is preceded (with few exceptions) by the 3rd pers. pl. clitic i: (56)

II, 61 58

(57)

II, 61 131-133

Quei che vend pû tant Those who sell-3SG then many imbratt ornaments “then, people that sell many ornaments” I becher ch’ hin the butchers who are pû i bon / then the good No i ghe NEG SCL.3PL IND.OBJ.CL fa tròp do.3SG too-much apiasè S’ a pleasure if a i tol via quel SCL.3PL take.3SG away what ch’ a ’s dè that a one must “Butchers—who are then good!—one does not do them a favour if one takes away from them what it is necessary”.

Table 6: Inventory of subject pronouns in Girolamo Maderno (Rabisch II, 61-II, 64)

1 2 3 4 5 6

Stressed pronouns

a

Subject clitics



a (o)

lu

a (o) a

e/i te l i

lor

a

i

Interrogative forms ho-i

A History of Personal Subject Pronouns in Milanese

159

5.4 Cheribizo: Isella (2005: 118-154); Morgana (2012: 64-65) Cheribizo is a poem consisting in 338 lines, written in a semi-popular variety of dialect. It was probably conceived inside Giovan Paolo Lomazzo’s Academy (see Isella, 2005: 121) and, according to Isella (2005: 124), the anonymous author is to be identified with Bernardo Rainoldi. Cheribizo is a poem in praise of the greatness of Milan, however, the poet did not want to use neither the normal dialect spoken in the city of Milan, nor the vernacular of Blenio Valley adopted by the other academicians: he preferred, instead, to make use of the artificial Bergamasque dialect already used in some comedies. This work is interesting for the history of the personal subject pronouns in Milan, because here we can find attestations of 2PL subject clitics (probably a spreading of a)24: (58)

vv. 9-10 S’ a If SCL.2PL corì run.IMP2PL volì want.2PL

volì want.2PL a Milan to Milan ach ben also well

ben well s’ if vestif dress

mangià eat.INF a SCL.2PL

6. Fabio Varese (1570-1630) Fabio Varese was a musician and a kind of “poète maudit”, author of Canzoni. He probably died during the plague which Manzoni has given an account of. In his language we can observe some instances of the cooccurrence of accented pronouns (or other DPs) together with correspondent clitics:25 (59)

II.1.12 Tì te favet la you SCL.2SG did.2SG the la coiona the idiot “You used to act the fool”

sempia e fool and

From the syntactic point of view, this sentence corresponds perfectly to the modern usage but, unlike the modern dialect, the co-occurrence of DP and scl is still optional in this period: 24 25

Numbers refer to Isella (2005: 127-154). Numbering refers to Stella et al. (1979) edition.

160

Chapter Six

(i) with scl: (60)

I.1.21

(61)

VIII.2

Ma “But quel who l’

quest el è nagott this scl is nothing” che fa l’ incognit that does the incognito è on coion SCL is an idiot “who plays at incognito is an idiot”

(ii) otherwise, also without scl: (62)

V.10

(63)

XI.9

quel moros Ø è un “That boyfriend Ø is a Quest non Ø this NEG Ø “This is not a madrigal”

pό poltron bit laggard” è madrigal is madrigal

6.1 Co-occurrence of SCLS and negation In this period, in case of co-occurrence of scl and negation, the reciprocal order still differs according to different persons: in case of 2nd pers. sing. there is still the old word order scl-neg-Verb: (64)

X.4

E mì sό te no and I know SCL.2SG NEG dis la veritá tell.2SG the truth “And I know you do not tell the truth”

On the contrary, in case of 3rd pers. sing., the new word order neg-scl-Verb is stable: (65)

XVIII.11

no

la

NEG

SCL.3SG REFL

s’

contenta is-satisfied

“She is not satisfied” So we can conclude that the process scl-neg-V > neg-scl-V was completed earlier for the 2nd than for the 3rd person.

A History of Personal Subject Pronouns in Milanese

161

Table 7: Inventory of subject pronouns in Fabio Varese Stressed pronouns 1 2 3 4 5 6

a

Subject clitics

Interrogative forms

mì tì lù / lé

a

e’ te, t’ el, l’ / la, l’

so-j vû-t fa-la

vù lor

a

i

7. Ambrogio Biffi: Prissian from Milan (1606) In 1606 Biffi published the “Prissian da Milan de la parnonzia milanesa”, which is an essay about the phonology of the Milanese dialect of his time, and probably the most important text written in a kind of scientific prose in Milanese literature. In this text we can still observe the use of the 1st sing. clitic e/i (which is very frequent, but sometimes for the 1st pers. sing. just the accented form mì might appear); like in other Northern Italian dialects, e/i can also be extended to 1PL: (66)

152.15 e and e

s if

i SCL.1SG

auess had.SUBJ1SG

vel to-you-it.CL fareu vedè made- COND.1SG see.INF “And if I had time, I would show it to you” 157.14 e se ben la schriuem and if well OBJ.CL write.1PL noma d’ ona sort e only of one way SCL.1SG vartirem notice.FUT1PL “And even if we write it just in one way, we will point out…” SCL.1SG

(67)

temp time

7.1 SCL al, ai = a + l, a + i SCL al is separable in /a + l/, e.g.:

162

(68)

Chapter Six

57

ch’ a nol chad sforzal that a NEG.SCL need force.OBJ.CL naghot nothing “that it does not need to force it at all”

However, when used as subject, the combination “a + l” is constant; also ai is separable in /a + i/, e.g.: (69)

56

A no i l’ an a NEG SCL.3PL OBJ.CL have.3PL foss mostrà tanc braù maybe showed so-many good schriciù writers “Did not so many good writers show that?”

7.2 Co-occurrence of scls with DPs When the subject is a preverbal DP there are two possibilities: (i) DP—a - SCL—inflected Verb: (70)

151.3-4 Quìj fiù d’ ingegn those boys of talent han comenzà […], have.3PL begun ven metènn to-you.CL put.PPLE.3PL chò head “Those canny types who began your mind…”

ch’ that ai a+SCL3PL in tel in the

[…], a-SCL.3PL put in

(ii) DP—Verb (without SCL): (71)

151.11 e an che i nost Ø se sìjen metù in vs “And even if ours Ø took the habit”

From these examples we can suppose that subject clitics can co-occur with a DP, which is presumably placed in TopP. We can also find at least

A History of Personal Subject Pronouns in Milanese

163

one case of pronominal reduplication co-occuring with an accented subject pronoun in Prissian for 3rd person plural: (72)

152.17 ma che lor ai but that they a+SCL.3PL l’an lechà inscì OBJ.CL have licked so “but that they have licked it this way”

SCL is present when the DP subject is postverbal, in case of an expletive SCL with impersonal verbs like “it seems”, “it needs”, “it is said”, with “there is”, with “to be” and “to have”, and with appositive clauses. However, in most of these cases, SCL is an optional and non-mandatory element. Moreover, like in Maderno’s poems, we find numerous instances of verb agreement in the singular with a 3rd person plural subject: this is still possible in the modern Milanese dialect, but in fewer contexts. In Prissian we can find this kind of agreement in: (i) interrogative and exclamative clauses: (73)

152.7

che sa i lor what know.3SG SCL.3PL they cosa sia on bel parlà? what be.SUBJ.3SG a fine speech “What does they know what a beautiful speech is?”

(ii) with subject postposed to the inflected verb: (74)

157.26 Es parnonzia come fava i one pronounce.3SG as did.3SG the Latin Latins “It is pronounced as the Latins did”

164

Chapter Six

Table 8: Inventory of subject pronouns in Prissian Stressed pronouns 1 2 3 4 5 6

a

Subject clitics

mi

e, i a

(a)l, la’ e

a

(a)i, i

nun lor

Interrogative forms

sa-i, a-i

8. Carlo Maria Maggi (1630-1699) As far as the morphological aspect is concerned, Maggi’s SCL system is still similar, where the attestations allow a comparison with, to Prissian’s one: The form al is still divisible into a + l:26 (75)

Ff II 271

(76)

Mm II 658

C’ al that a+SCL3SG senta Meneghin hear.SUBJ3SG Meneghin s’el fa conzett if SCL3SG does reasoning “May you listen to Meneghin, if he is reasoning well” C’ a ne la se that a NEG SCL.3SG REFL dubitta doubt.3SG “That she may not doubt”

However, the combination of a + i for the 3rd person plural does not occur anymore; in those cases where the 3rd person plural. SCL occurs, its form is i, or sometimes only a occurs:

26

Numbering here refers to comedy, act and verse, as per Isella (1964) edition.

A History of Personal Subject Pronouns in Milanese

(77)

Bb Pr II 53

(78)

Mm III 975

165

Da mett pagura ai fang [to put fear] to-the children quand i se stinnen when SCL.3PL REFL insist.3PL “To frighten the children when they (SCL.3PL + REFL) insist” A me pæren a to-me.CL seem.3PL prodezz da biridoeù feats of thoughtless “They seem to me the exploits of heedless people”

8.1 Co-occurence with DPs (1) When the DP subject occurs before the verb, subject clitics have an anaphoric function: (79)

Mm Pr II 35-36 La vedeva l’ è come the widow SCL.3SG is like la gallina, / semper la the hen, always SCL.3FSG ruspa, e semper la pokes and always SCL.3FSG rangogna complains “The widow is like the hen, she is always poking around and always complaining” When the subject of the verb is a DP there are two possibilities: (i) DP—scl—verb

(80)

Cm III 634

(81)

Ff I 463-4

I paroll d’i pastogg i corren via “The words of the tales SCL.3PL run away” La tosa sbaguttì/ the girl dumbfounded/ la vegnè in volt SCL.3FSG turned in her face “Dumbfounded, the girl’s face turned (like a brazier of fire)”

166

Chapter Six

(ii) DP—verb (without SCL) (82)

Bb I 285 Quand el sò spenditor / Ø compær in su’l Verzé

(83)

Mm III 910

“When her shopping-man / Ø appears at the market” Ma el valor Ø è prudent “But (the) valour Ø is prudent”

Subject clitics repeated in coordination are very frequent: (84)

Mm II 463

Subet al streng suddenly a+SCL.3SG closes i ogg e al the eyes and a+SCL.3SG se stremiss REFL frightens “He suddenly closes his eyes and gets frightened”

From these examples we can conclude that DP can still be in complementary distribution with SCL. If SCL and DP co-occur, probably DP is in TopP. (2) When the DP subject occurs after the verb, we still have two possibilities: (i) SCL—verb—DP (85)

Ff I 706 Quand al ven San Miché when a+SCL.3SG comes Saint Michael “When St. Michael comes (to pay the rent)” (ii) verb—DP (without SCL)

(86)

Cc 710 d’onde ven Ø sta tremenda stravasciǽ?

(87)

Cm Int I 93

“whence comes Ø this terrible collapse?” Ø Hin i simbij de tugg i prum usanz “They are the apes of all the first customs”

8.2 Co-occurrence of SCLs with NEG Subject clitics are firmly placed for all persons between preverbal negation and the inflected verb:

A History of Personal Subject Pronouns in Milanese

(88)

Mm Pr II 173

(89)

Mm Pr II 60

(90)

Mm Pr II 49

(91)

Mm Pr II 77

167

Tì, che no t’ you, who NEG SCL.2SG hé volsù have.2SG wanted “You, who did not even want (to bear)” No t’ hé NEG SCL.2SG have.2SG nagott de bon nothing of good “You have nothing good (but the clack)” che s’ al s’ that if a+SCL.3SG REFL intedarà no ’l understand.FUT NEG SCL.3SG sarà pocco be.FUT little “That if this is understood it will be no small matter” Che no ’l po’ god that NEG SCL.3SG can enjoy nagott nothing “That he can’t enjoy anything”

9. Cherubini’s notes (1856) As an appendix to the 5th volume of Milanese—Italian Dictionary in the essay “Nozioni filologiche intorno al dialetto milanese”, Cherubini makes some interesting observations about changes that occurred in Milanese dialect between the 18th and 19th century: (i) at the end of the 18th century the 3PL SCL i went out of use, but was still being used in peripheral areas during Cherubini’s time (nowadays it is present in some dialects apart from Milanese). (ii) he considers verbal forms without subject pronoun elements as non grammatical, such as corri (I-run), corret (you-run), corr (she/heruns); córrem (we-run), corrii (you-run), corren (they-run), all of these being replaced by: mi corri, ti te corret, lu el corr; nun correm, violter corrii, lor corren. He notices that, while until the end of the 18th century Biffi could say El natural Ø sporsg squas semper a tœù i cos par el so drizz (“Natural things almost always appear in the right way”) and Maggi

168

Chapter Six

could say: Mi ghen doo vintott sold, lu Ø se reffigna (“I give him twenty eight coins and he…”), in the 19th century, Cherubini must say: “El natural el sporsg, etc.”; “Lu el se reffigna, etc.”, i.e. with the obligatory presence of the 3SG SCL. (iii) Cherubini still notes interrogative forms like FornireT?, forniraL? (“will-finish 2SG.SCL?”, “will-finish 3SG.SCL?”), while in the same period peripheral dialects (e.g. Brianza dialects) extended this peculiarity also to 3rd person plural: Fornira-i?, “finish-they?”; E-i sœu quij fiœu lì?, “is-they yours those children there?”. In Cherubini’s opinion, these idioms were “denied to Milanese by the nature of its dialect”, however, contrary to his opinion, these forms used to be part of the urban inventory during the 17th century.

10. The fate of clitic a in contemporary Milanese The clitic a has not completely disappeared in contemporary Milanese: I could record it used by some speakers in Vai (1999). The clitic a is still present in existential constructions, e.g. it can appear optionally as 3PL clitic: lur (a) gh’an “they (a) have”. It can be used together with other clitic pronouns in certain exclamative sentences, e.g.: (cum’è) a te se mövet?! and: a gh’è rivà l Mario! “Mario has arrived”. In this last case, with a postposed singular or plural subject; this use is attested for other Northern Italian dialects as well (see Benincà 1994: 25-26). Further research is needed in order to clarify the syntax of the clitic a and above all to clarify whether it corresponds to a single form or (at least two) different homophonic shapes: one corresponding to the clitic a studied by Benincà, and the other being a substitute of the older SCL.3PL i.27

11. Conclusions The phenomena surveyed in the present article lead to the following conclusions: (i) Bonvesin’s pronominal system already shows some instances of pronominal reduplication, a phenomenon which is linked to the origin of the SCL system; (ii) in Milanese the change SCL—NEG—V > NEG—SCL—V begins at the end of the 14th century and ends in the 17th century; 27

As for the Ticinese dialect, Sganzini (VSI, vol. I, p. 5) noticed: “Nella 3 a pers. pl. a va penetrando per influsso del lombardo comune, soprattutto nei centri e nelle regioni a sud del Ceresio”.

A History of Personal Subject Pronouns in Milanese

169

(iii) SCLs can be in complementary distribution with DPs at least up to the end of 17th century; (iv) from the end of the 18th century the system of obligatory scls is reduced to the 2nd and 3rd persons singular; a clitic a can replace (although not obligatory) the old 3PL SCL i. (v) at least from the 18th century SCLs become obligatory (as also stated by Cherubini); (vi) in Delio Tessa’s poems (Isella: 1985) there are still cases of interrogative inversion, e.g. Isella (1985: 33): dov’ell? “where is it (SCL)?”; Isella (1985: 125): Come valla? “how is it (SCL)?”; (Isella 1985: 338): In dov’eel mo? “where is he (SCL) now?”, which now have disappeared.

Corpus AIS = Jaberg, Karl—Jud, Jakob, Sprach- und Sachatlas Italiens und der Südschweiz, Zofingen 1928-1940.

References Benincà, P., 1994, La variazione sintattica, Bologna, il Mulino. Benincà, P., 2004, The Left Periphery of Medieval Romance, downloaded from http://www.humnet.unipi.it/slifo/2004vol2/Beninca2004.pdf. Brandi, L., P. Cordin, 1981, Dialetti e italiano: un confronto sul Parametro del Soggetto Nullo, Rivista di Grammatica Generativa, 6, 3–32. Cherubini, F., 1856, Vocabolario Milanese-Italiano, vol. V. Sopragiunta. Nozioni filologiche intorno al dialetto milanese. Saggio d'osservazioni su l'idioma brianzuolo, suddialetto del milanese, Milano, Dalla Società tipografica de’ classici italiani. Colombo, M., 2016, Passione Trivulziana. Armonia evangelica volgarizzata in milanese antico, Berlin/Boston, Walter de Gruyter. Contini, G.,1941, Opere volgari di Bonvesin dra Riva, Roma, Presso la Società. Isella, D., 1964, Carlo Maria Maggi. Il teatro milanese, 2vv, Torino, Einaudi. Isella, D., 1975, (a cura di) Carlo Salvioni. Fonetica e morfologia del dialetto milanese, Pacini. Isella, D.,1979, Lo sperimentalismo dialettale di Lancino Curzio e compagni, in: F. Alessio, S., Angelo (eds), In ricordo di Cesare Angelini, Milano, il Saggiatore, 147–159. Isella, D. 1985 (ed.) Delio Tessa. L’è el dì di Mort, alegher! De là del mur e altre liriche, Torino, Einaudi. Isella, D., 1993 (ed.), Giovan Paolo Lomazzo e i Facchini della Val di Blenio. Rabisch, Torino, Einaudi. Lepschy, G. C.,1965, Una fonologia milanese del 1606: il Prissian da Milan della Parnonzia Milanesa, L’Italia dialettale 28, 143–180.

170

Chapter Six

Levi, U. (1904) I monumenti del dialetto di Lio Mazor, Venezia, Visentini cav. Federico. Loporcaro, M., L. Pescia, R. Broggini, P. Vecchio, 2008 (eds), Carlo Salvioni. Scritti linguistici, 5vv, Bellinzona, Edizioni dello Stato del Canton Ticino. Lorck, J. E.,1893, Altbergamaskische Sprachdenkmäler, Halle, Verlag von Max Niemeyer. Manzini, M. R., L. Savoia, 2005, I dialetti italiani e romanci. Morfosintassi generativa, vol. I, Alessandria, Edizioni dell’Orso. Meyer-lübke, W.,1897, Zur Stellung der tonlosen Objektspronomina, Zeitschrift für romanische Philologie, 21, 313–334. Morgana, S., 2012, Storia linguistica di Milano, Roma, Carocci. Pires, A., 2006, The Minimalist Syntax of Defective Domains: Gerunds and Infinitives, Amsterdam/Philadelphia, John Benjamins. Poletto, C.,1995, The Diachronic Development of Subject Clitics in North Eastern Italian Dialects, in: A. Battye, I. Roberts (eds), Clause Structure and Language Change, Oxford University Press, Oxford-New York, 295–334. Poletto, C.&Zanuttini, R. (2010) Sentential Particles and Remnant Movement, in Benincà, P.&Munaro, N. (eds.) Mapping the Left Periphery. The Cartography of Syntactic Structures, Volume 5, Oxford/New York, Oxford University Press, 201-227. Rohlfs, G., 1966–1969, Grammatica storica della lingua italiana e dei suoi dialetti, 3vv, Torino, Einaudi. Rizzi, L., 1986, On the Status of Subject Clitics in Romance, in: O. Jaeggli, C.Silva-Corvalán (eds), Studies in Romance Linguistics, Foris, Dordrecht. Roberta, I., 2007, Diachronic Syntax, OUP, Oxford/New York. Salvi, G., 2004, La formazione della struttura di frase romanza, Tübingen, Max Niemeyer Verlag. Sanga, G., 1984, Dialettologia lombarda, Dip. di Scienza della Letteratura, Università di Pavia, Aurora Edizioni, Pavia. Stella, A.- B., Massimo-Marchi, R.,1979, (eds), Fabio Varese. Canzoni, Milano, All’insegna del pesce d’oro (Scheiwiller). Vai, M.,1996, Per una storia della negazione in milanese in comparazione con altre varietà altoitaliane, Annali della Facoltà di Studi Umanistici dell'Università degli Studi di Milano, XLIX, fasc. I (gennaio-aprile), 57–98. Vai, M., 1999, “Note sulla frase esclamativa nel dialetto milanese”, Quaderni di lavoro ASIt - ASIt Working Papers, 3, http://asit.maldura.unipd.it/papers.html. Vanelli, L., 1998, I dialetti italiani settentrionali nel panorama romanzo, Roma, Bulzoni. Wackernagel, J., 1892, “Über ein Gesetz der indogermanischen Wortstellung”, Indogermanische Forschungen, 1, 333–436. Wilhelm, R., F. de Monte, M. Wittum, 2011, Tradizioni testuali e tradizioni linguistiche nella Margarita lombarda, Heidelberg, Universitätsverlag. Wilhelm, R., 2006, Bonvesin da la Riva. La Vita di Sant’Alessio, Tübingen, Max Niemeyer Verlag.

CHAPTER SEVEN ON THE ETHICAL DATIVE IN OLD AND MODERN ROMANIAN MIHAELA TĂNASE-DOGARU, CAMELIA UȘURELU University of Bucharest [email protected]; [email protected]

1. Introduction The focus of the present paper is represented by non-core argument datives of the type in (1): (1)

a.

b.

Mi-l ia odată I.DAT.CL-he.ACC.CL takes once Făt-Frumos și mi Prince Charming and I.DAT.CL ți-l vâră pe zmeu you.DAT.CL-he.ACC.CL thrust DOM ogre în noroi până la glezne (in GALR: 207) in mud up to ankles “Prince Charming falls upon the ogre and thrusts him in the mud up to his ankles” Je t’ acheterais un I you.DAT.CL would.buy a cadeau a Pierre present to Pierre “I tell ya, I would buy Peter a present” (Sportiche 1993: 18, quoted in Diaconescu 2004: 229)

172

Chapter Seven

c.

Figlio, chi me t’ha son, who me.CL.1SG you.CL.2SG-has morto? killed? “Son, who killed you?” (Jacopone da Todi, Laude XCIII, quoted in Cuervo: 2003)

Following GALR (2005: 207), RGR (2013: 242), Roberge and Troberg (2009), Jouitteau and Rezac (2008), Diaconescu (2004), we will call such constructions ‘ethical datives’, in order to avoid confusion with other non-core datives, such as the dativus commodi/incommodi. Different labels have been used more or less interchangeably in the literature to refer to Romance dativus commodi/incommodi and/or dativus ethicus: “ethical”, “affected”, “free”, “nonlexical dative, or “dative of interest”. ‘Ethical datives’ are traditionally assumed to be optional and to indicate the affective/emotional involvement of the speaker (and hearer) in the narrated events (GALR: 207, RGR: 242). Pragmatically, ethical datives are assumed to have the persuasion role (GALR: 207). Also, ethical datives are assumed to characterize older stages of Romanian, being highly infrequent in Modern Romanian, their distribution being limited to fixed expressions (2): (2)

Pe unde-mi umbli? (RGR: 242) Over where-I.DAT.CL walk.2.SG? “Where are you wandering about?”

According to Irimia (2000: 108), ethical datives are characteristic of the familiar, regional and colloquial register. In Diaconescu (2004: 226), it is claimed that ethical datives can appear with predicates of all types and, as a result, they can be ambiguous between an ethical dative or other types of datives; to avoid ambiguity, researchers narrow down the range of ethical datives to first and second person datives, which express ‘the interest of the person who narrates or datives which are soliciting the interest of the person who reads/listens to the events’ (Diaconescu 2004: 226). The paper investigates the syntax of ethical datives in Old Romanian, while also amassing a small corpus; it also shows that Modern Romanian registers frequent usage of the ethical dative in motherese.

On the Ethical Dative in Old and Modern Romanian

173

2. The corpus1 Our Old Romanian corpus registers both examples from the 17th century (3) and the 19th century (4): (3)

a.

b.

1

Alexandre, fie voia ta deplin Alexandru.VOC, be will.DEF your whole și mi te du cu and I.DAT.CLyou.REFL.CL go with ajutorul lui Amon! help.DEF of Amon! “Alexander, do as you wish and leave with the help of Amon!” (Alexandria: 19) Carele împreună mi who.DEF together I.DAT.CL l-ai îndulcitu-mi he.CL-have.2SG sweetened-I.DAT la bucate, la casa lui at food, at house.DEF of Dumnedzău îmblam God walk.I.IMPERF.1PL “The one whom you enticed to food (= eternal life) and I walk together to the house of God” (Dosoftei, Psaltirea de-nțăles: 413)

In the 16th and 17th centuries, there are very few attestations of ethical datives. There is no attestation of the structure in the following works, which are quoted in the Sources section: Antim Ivireanul (Didahii), Ioan Cantacuzino (Patru apologii pentru religia creştină şi patru oraţii traduse în limba română la mijlocul secolului al XVII-lea de Nicolae Spătarul (Milescu)), Dimitrie Cantemir (Istoria ieroglifică, Divanul), Coresi (Tâlcul evangheliilor, Molitevnic rumânesc, Carte cu învăţătura), Miron Costin (Letopiseţul Ţărîi Moldovei), Cristina-Ioana Dima (Apocalipsul Maicii Domnului), Documente și însemnări românești din secolul al XVI-lea, Dosoftei (Psaltirea în versuri), Evanghelie învăţătoare, Floarea darurilor, Texte româneşti din secolul al XVI-lea, Istoria Ţării Româneşti de la octombrie 1688 până la martie 1717, Istoria Ţării Româneşti de stolnicul Constantin Cantacuzino, Liturghierul lui Coresi, Manuscrisul de la Ieud, Ion Neculce (Letopiseţul Ţării Moldovei), Palia de la Orăştie, Pravila ritorului Lucaci, Psaltirea Hurmuzaki, Sindipa, Grigore Ureche (Letopiseţul Ţării Moldovei), Varlaam (Răspunsul împotriva Catihismusului Calvinesc), Varlaam şi Ioasaf.

174

Chapter Seven

c.

(4)

a.

b.

c.

Post şi sărbătoare şi fast and celebration and praznicile voastre le ureaşte feast.PL.DEF your them hates sufletul mieu, mi vă feacet de soul.DEF my I.DAT.CLyou.ACC make of saţâu! disgust “My soul hates your fasts and celebrations and feasts, you become disgusting” (Dosoftei, Parimiile preste an: 105) Atunci Ipate odată mi then Ipate once I.DAT.CL ț-o și înșfacă de you.DAT.CL-she.ACC.CL and grab.3SG of cozi, o trântește la braids, she.ACC.CL knock.3.SG at pământ ș-o ține ground and-she.ACC.CL hold.3.SG bine. well “Then Ipate grabs her braids, knocks her down and holds her tight” (Creangă, Povești: 82) Ei, că bine mi well, that well I.DAT.CL te-am căptușit you.ACC-have.1.SG caught “Well, I have caught you good!” (Creangă, Povești: 96) când văd că mâța face mărazuri, when see.1.SG that cat.DEF makes whims ți-o strâng de you.DAT.CL-she.ACC.CL squeeze of coadă, de mănâncă mere pădurețe tail, that eat.3SG apples sour “When I see that the cat is whimsical, I squeeze its tail and it ends up eating sour apples” (Creangă, Povești: 108)

On the Ethical Dative in Old and Modern Romanian

d.

e.

f.

175

începe Flămânzilă a start.3SG Flămânzilă to cărăbăni deodată în gură câte fill suddenly in mouth DISTR o haraba de pâine și câte a cart of bread and DISTR o ialoviță întreagă, și răpede a cow whole, and quickly mi ți le-a I.DAT.CL you.DAT.CLthem.ACC-have.3SG înfulicat. gobbled “Flămânzilă suddenly starts filling his mouth with carts of bread and whole cows and gobbles them quickly!” (Creangă, Povești: 124) și odată mi ți-l and once I.DAT.CLyou.DAT.CL-he.ACC înșfacă cu dinții de grab.3.SG with teeth.DEF of cap head “And suddenly he grips his head with his teeth” (Creangă, Povești: 134) și odată mi ț-o and once I.DAT.CLyou.DAT.CL-she.ACC înșfacă ei, unul de o snatch they.NOM one.DEF of a mână și altul de cealaltă hand and other.DEF of another “And suddenly they grab her, one of them by a hand and the other one—by the other hand” (Creangă, Povești: 128)

176

Chapter Seven

g. .

Apoi, cuprinzând then, seizing brațe, găbuiește, arms find.3.SG ț-o you.DAT.CL-she.ACC coadă tail “Then, taking the moon bird and grabs its tail…”

luna în moon-the in păsărica, mi bird.DEF I.DAT.CL înșfacă de grab.3.SG of

in his arms, he finds the little (Creangă, Povești: 128)

All the examples in (4) contain examples of ‘non-actantial’ datives (Delbecque & Lamiroy 1996: 106-107), in the sense that they are not part of the valency of the verb but having an expressive function and grounding ‘the event structure in relation to the speech participants’. Our Modern Romanian corpus contains examples such as those in (5), a lot less ‘literary’ but showing the same kind of emotional involvement. (5)

a.

b.

c.

2

Dar azi mi-a făcut but today I.DAT.CL-have.3.SG made febră dimineața.2 fever morning “But today he developed a fever in the morning” Mi-a mâncat azi doar 50 I.DAT-have.3.SG eaten today only 50 de grame de morcov3 of grams of carrot. “Today he has eaten only 50 grams of carrot.” Dar de mâncat îți but of eating you.DAT.CL mănâncă?4 eat.3SG? “As for eating, does he eat (enough)?”

http://doctoritadecopii.ro/?p=2388. http://www.romedic.ro/forum/crestere-in-greutate-cu-probleme-10491. 4 Fragment of conversation, May 2014. 3

On the Ethical Dative in Old and Modern Romanian

d.

e.

177

Sunt supărată că nu be.1.SG upset.FEM that not mi-a învățat poezia.5 I.DAT.CL- have.3.SG learnt poem.DEF “I am upset because he has not learnt the poem” mi l-a luat I.DAT.CL he.ACC-have.3.SG taken pe copilul ăla și DOM child.DEF and i-a dat una he.DAT.CL-have.3.SG given one în fața mea...6 in face.DEF my “She took that child and slapped him in front of me”

According to RGR (2013), ethical dative clitics can co-occur with indirect object dative clitics and are placed in front of the indirect object dative; two ethical datives can co-occur within the same clitic cluster, in the order ethical dative.1SG > ethical dative.2SG > dative clitic indirect object (RGR 2013: 257): (6)

a.

b.

c.

Mi-i aduce I.DAT.CL-CL.3SG.DAT brings împăratului merele. emperor.DEF.DAT apple.PL.DEF Ți-i aduce you.DAT.CL-CL.3SG.DAT brings împăratului merele. emperor.DEF.DAT apple.PL.DEF Mi ți-i aduce I.DAT.CLyou.DAT.CL-CL.3SG.DAT brings împăratului merele. emperor.DEF.DAT apple.PL.DEF

Ethical datives can co-occur with direct object accusative clitics, in the order ethical dative.1SG > ethical dative.2.SG > accusative clitic direct object (RGR 2013: 257):

5 6

Fragment of conversation, October 2016. Fragment of conversation, September 2016.

178

(7)

Chapter Seven

a.

b.

c.

Mi-l aruncă I.DAT.CL-CL.3SG.MASC.ACC threw.3SG vrăjitoarea peste șapte codri! witch.DEF over seven woods Ți-l aruncă you.DAT.CL-CL.3SG.MASC.ACC threw.3SG vrăjitoarea peste șapte codri! witch.DEF over seven woods Mi ți-l I.DAT.CLyou.DAT.CL-CL.3SG.MASC.ACC aruncă vrăjitoarea peste șapte witch.DEF over seven threw.3SG codri! woods

In clitic clusters containing an ethical dative, direct object accusative clitics are in complementary distribution with indirect object dative clitics (RGR: 258). Our corpus features mainly first person ethical dative clitics; a second well-represented case is first followed by second person ethical dative clitics; clitic clusters containing both first and second person ethical datives and accusative clitics are rarer. The Modern Romanian examples only feature distinct cases of ethical datives in the first person, and distinct cases of second person ethical datives; the lack of ethical dative clitic clusters may be attributed to the oral nature of the register.

3. Analysis In analyzing ethical datives we start from the assumption that they have (at least) two disambiguating syntactic features (see Cuervo 2003, Diaconescu 2004, Roberge and Troberg 2009, Jouitteau and Rezac 2008): (a) ethical datives are restricted to 1st and 2nd person; (b) ethical datives do not allow a full pronominal DP corresponding to the benefactive clitic. (8)

a.

Unde mi-ai umblat where me.CL.DAT-have(you) walked (*mie)? (me.DAT) “Where were you?” (Diaconescu 2004: 227)

On the Ethical Dative in Old and Modern Romanian

b.

179

Me le dieron un me.DAT.CL him.DAT.CL gave an helado al nino ice-cream to kid (*a mi) (Cuervo 2003: 195) (me.DAT) “They gave the kid an ice-cream on me”

Starting from an analysis of the dativus commodi/incommodi, Roberge and Troberg (2009) maintain a clear distinction between the dativus commodi/incommodi and the dativus ethicus and establish a list of properties that differentiate the ethical dative from other types of datives. According to Roberge and Troberg (2009), ethical datives are distinguished by means of several identifying properties. Firstly, despite the similarity between the DCI and the ethical dative, in that neither is subcategorized by the verb, the ethical dative is limited to first and second person pronouns. Ethical datives are adjuncts (see also Franco and Huidobro 2008). Rather than denoting the person who benefits or who is adversely affected by the event (which is the case of the dativus commodi/incommodi), the ethical dative “designates, in emphatic clauses, the person taken as witness among the performers of the utterance” (Roberge and Troberg 2009: 255). In other words, the ethical dative always has a non-referential reading. Cardinaletti and Starke (1994: 51) call the ethical dative a discourse particle, in the sense that “there is no referent to these pronouns, not even derivatively.” (9)

Quid mihi Celsus agit? what I.DAT Celsus.NOM act.3SG “How, pray7, doth Celsus fare?” (Horace, Epistulae 1,3,15, in Roberge and Troberg 2009: 255; Woodcock 1959: 47, quoted in Roberge and Troberg 2009)

Thirdly, ethical datives are not arguments; they do not affect the truth conditional meaning of the sentence; they invoke the speaker or the 7

Roberge and Troberg (2009: 255) note that Woodcock (1959) translates the first person pronouns of the Latin DE as ‘please!’ or ‘pray’, and the second person pronoun as ‘for your pleasure I tell you this’, ‘you will be surprised to hear’, or ‘lo and behold!’. These translations reflect the non-referential reading of the ethical dative.

180

Chapter Seven

addressee as witness or vaguely affected party (Jouitteau and Rezac 2008: 98; see also Franco and Huidobro 2008). In Jouitteau and Rezac (2008) for French, and in Diaconescu (2004) for Romanian, it is argued that the ethical dative is not only semantically distinct from the whole range of core and ‘extended’ datives (datives of inalienable possession, benefactive datives, affected datives) but also syntactically distinct. Jouitteau and Rezac (2007, 2008) work out a number of syntactic tests to distinguish between ethical datives, like the ones in (10) and the rest of the dative domain: (10)

Je te me vais te me vous I 2SG 1SG go 2SG 1SG 2PL lui faire passer un sale quart 3SG.DAT.CL make pass a dirty quarter d’heure -hour “I’m gonna make him spend a lousy quarter-hour…”

The tests are drawn up in order to single out ethical datives according to formal criteria, since meaning is not a sure criterion for ethical datives; “the meaning of extended datives, particularly benefactive and affected, shades into that of the ethical dative” (Jouitteau and Rezac 2008: 98).

3.1 Restriction to first and second person and compatibility with other datives Argumental dative clitics cannot co-occur; but they can co-occur with ethical datives. Ethical datives are compatible with another dative (lexical, non-lexical or ethical). (11)

a.

Elle (me) lui She 1SG 3SG.DAT.CL un bébé dans les a baby in the “She has put a child in his arms”

a has bras. arms

mis put

On the Ethical Dative in Old and Modern Romanian

b.

(11’)

a.

b.

181

Elle (te/te me/te she 2SG/2SG 1SG/2SG me/nous/*leur) lui a 1SG/1PL/*3PL.DAT.CL 3SG.DAT.CL has attrapé trois rhumes cet hiver caught three colds this winter “She caught three colds this winter on her, you know” Și unde nu mi and where not 1SG i-a turnat un copil he.DAT.CL-have.3SG made a child repede... quickly “She was very quick in having a child with him” Mi/ți/ ne-a 1SG/ 2SG/ 1PL.DAT.CL-have.3SG răcit de trei ori până caught-a-cold of three times so acum far “He has caught a cold three times on me / you / us”

As for the restriction to first and second person, there is a longstanding tradition of looking upon third person dative clitics as indicating the possessive dative or the benefactive dative, especially if the object is definite: (12)

a.

I-a

învățat poezia. learnt poem.DEF “She/he has learnt (her) poem” Mi-a învățat poezia. CL.1SG.DAT-has learnt poem.DEF “She has learnt the poem (for me)” CL.3SG.DAT-has

b.

This restriction has to do with interpreting the relation between the dative clitic and the accusative object as one of possession. In other words, in (12a), if there is a possession relation between the poem (as possessee) and the dative (as possessor), the clitic is interpreted as a dative of possession. In a larger scenario, one in which there is an indirect discourse source for (12a) and there is no possessive relation between the dative clitic and

182

Chapter Seven

the accusative object, the dative clitic acquires an ‘ethical’ reading. The examples in (13) are such ambiguous cases: (13)

a.

b.

c.

I-a fugit de la școală. 3SG.DAT.CL-has run of at school “(Her child) has run away from school” I-a lipsit de la cursuri. 3SG.DAT.CL-has missed of at courses. “(Her child) has skipped classes” Copilul îi merge Child.DEF 3SG.DAT.CL walks deja (Niculescu 2008: 497) already “Her child already walks”

Diaconescu (2004) claims that examples like (13) are high applicative arguments, where there is no possession relation but a benefactive relation between the event and the dative: (14)

a.

b.

Mihaela îi aleargă antrenorului Mihaela he.DAT.CL run trainer.DEF.DAT o jumătate de a half of oră. (Diaconescu 2004: 218) hour “Mihaela runs half an hour for the trainer” Mihaela i-a alergat antrenorului Mihaela he.DAT.CL-has run trainer.DEF.DAT o cursă întreagă. a race full “Mihaela has run a full race for the trainer”

The syntactic structure proposed in Diaconescu (2004) follows Pylkkänen’s (2002) distinction between the high applicative head and the low applicative head:

On the Ethical Dative in Old and Modern Romanian

(15) Voice0

183

VoiceP 2 ApplP 2

DPDAT antrenorului mie ție

Appl’ 2 Appl0 vPACT îi 2 v0 2 Root DPObject alergo cursă

The hypothesis we may formulate is therefore that some ethical datives are ‘derived ethical datives’; they are high applicative ‘real’ dative arguments acquiring ‘ethical’ readings. This is in line with Cuervo (2003), who proposes that ethical dative clitics are similar to high applicative clitics. 3.2. Invisibility to the ‘Person Case Constraint’ According to the “Person Case Constraint”, a first and second person accusative clitic is blocked in the presence of a dative clitic; ethical datives are invisible to the constraint. This most likely happens because ethical datives are generated outside the case system. Albizu (1997) proposes that ethical datives are generated above T (see also Rivero 2003). (16)

a.

b.

(17)

a.

b.

*Elle vous m’ a trouvé she 2PL 1SG has found “She has found you for me / me for you” Demain je (me) vous (me) tomorrow I 1SG 2PL.ACC 1SG emmène en vacances take1.SG in vacations “Tomorrow I will take you on vacation” *Ea te m-a găsit She 2SG 1SG-has found “She has found you for me / me for you” Mâine mi te tomorrow 1SG you.SG.ACC iau în vacanță take.1.SG in vacation “Tomorrow I will take you on vacation”

184

Chapter Seven

3.3 Clitic only There is no independent realization by a DP or a PP for ethical datives. Jouitteau and Rezac (2008) suggest that this has to do with case— if ethical datives are generated outside the thematic and case domains, they couldn’t be DPs because they couldn’t satisfy case requirements. (18)

*Mi l-a luat de 1SG he.ACC.CL-has taken of mamei mother.DEF.DAT “She has grabbed his hair on me”

păr hair

Diaconescu (2004) explains the impossibility of doubling ethical datives by a full pronominal DP by the fact they are not ordinary high applicatives but high applicative projections above Tense: (19) Appl0 mi-/ţi-

ApplP 2

Diaconescu (2004: 228) TP 2 VoiceP 2 DPSubj pro

Voice’ 2 Voice0 vPACT 2 v0 Root umbl-; lupt-

3.4 Ethical datives may appear in imperatives Jouitteau and Rezac (2008) analyze the possibility of French ethical datives to appear in embedded clauses (infinitives included), in questions and imperatives as an indication of the fact that ethical datives are independent of the properties of the C-system (Force, Mood, Finiteness) suggesting a lower position. (20)

Mi te 1SG 2SG.REFL “Look here!”

uită look

aici! here!

On the Ethical Dative in Old and Modern Romanian

185

3.5 No ‘tail effect’ In French affected datives are illicit or degraded with VPs lacking internal arguments; ethical datives share this restriction but are licit with any adjunct. In Romanian, ethical datives do not seem to show ‘tail effects’. (21)

(22)

Alfred m’ a roté *(pour Alfred 1SG has burped for choquer ses invités to shock his guests “Alfred burped to shock his guests” Copilul mi-a strănutat. child.DEF 1 SG-has sneezed “The child has sneezed on me” 3.6 Dative clitics and focus

The structure in (18), proposed by Diaconescu (2004) for Romanian ethical datives, suggests that this subtype of applicative need not project a specifier in the same way as ordinary high applicatives that are expressed only by a dative clitic. Jaeggli (1982) presents ethical or benefactive clitics as a case of clitics that are not generated in object position (see also Franco and Huidobro 2008). Since ethical dative clitics do ‘not alternate with any other postverbal object position, we can use this as strong evidence that the clitic is generated by the base in clitic position’(Jaeggli 1982: 18). Diaconescu (2004) follows Jaeggli’s proposal that the ethical dative is not in argument position and argues that, because ethical datives do not bear any relation to the subject or the activity/state expressed by the verb, ethical datives are generated in the higher part of the syntactic structure of a sentence to check the [+Focus] feature (Diaconescu 2004: 229).

4. Conclusions The present paper has focused on the syntax and interpretation of ethical datives in Old and Modern Romanian. The conclusions it has reached are the following. First, ethical datives are not extinct in Modern Romanian; they seem to be highly frequent (at least) in a sub-sociolect of Modern Romanian, where the dative clitic is not interpreted as a possessor or as a benefactive. Secondly, ethical datives are a sub-type of high

186

Chapter Seven

applicative phrases, projected above Tense, in the higher syntactic structure of a sentence, checking a [+Focus] feature. Thirdly, there are two sub-types of ethical datives: standard (1st and 2nd person pronouns) and ‘derived’ (3rd person pronouns acquiring ‘ethical’ readings in the absence of a possession/benefactive relation.

Corpus Alexandria (Istoria despre marele Alexandru-împărat), in Cărţile populare în literatura românească, ed. by Ion Chiţimia & Dan Simionescu, Bucharest, Editura pentru Literatură, 1963. Antim Ivireanul, Didahii, in Opere, ed. by Gabriel Ştrempel, Bucharest, Editura Minerva, 1972. Cantacuzino, Ioan, Patru apologii pentru religia creştină şi patru oraţii traduse în limba română la mijlocul secolului al XVII-lea de Nicolae Spătarul (Milescu), ed. by Eugenia Dima, Iași, Editura Universităţii „Al. I. Cuza” din Iaşi, 2011. Cantemir, Dimitrie, Divanul sau Gâlceava înţeleptului cu lumea, ed. by Virgil Cândea, Bucharest, Editura Minerva, 1990. Cantemir, Dimitrie, Istoria ieroglifică, ed. by P.P. Panaitescu & Ion Verdeş, Bucharest, Editura Minerva, 1983. Coresi, Tâlcul evangheliilor (=Cazania I) şi Molitevnic rumânesc [1567-1568], ed. by Vladimir Drimba, Bucharest, Editura Academiei Române, 1998. Costin, Miron, Letopiseţul Ţărîi Moldovei, in Opere, ed. by P. P. Panaitescu, Bucharest, Editura de Stat pentru Literatură şi Artă, 1958. Creangă, Ion, Povești, amintiri, povestiri, ed. by Iorgu Iordan & Elisabeta Brâncuș, Bucharest, Editura Minerva, 2015. Diaconul Coresi, Carte cu învăţătura (1581), ed. by Sextil Puşcariu, Alexie Procopovici, Bucharest, Atelierele Grafice Socec & Co., 1914. Dima, Cristina-Ioana, Apocalipsul Maicii Domnului. Versiuni româneşti din secolele al XVI-lea—al XIX-lea, Bucharest, Editura Academiei Române, 2012. Documente și însemnări românești din secolul al XVI-lea, ed. by Gheorghe Chivu, Magdalena Georgescu, Magdalena Ioniță, Alexandru Mareș & Alexandra Roman-Moraru, Bucharest, Editura Academiei Române, 1979. Dosoftei, Psaltirea în versuri (1673), ed. by N.A. Ursu, Iaşi, Mitropolia Moldovei şi Sucevei, 1974. Dosoftei, Psaltirea de-nţăles [1680], ed. by Mihaela Cobzaru, Iaşi, Casa Editorială „Demiug”, 2007. Dosoftei, Paremiile preste an, Iaşi, 1683, ed. by Mădălina Ungureanu, Iaşi, Editura Universităţii „Alexandru Ioan Cuza”, 2012. Evanghelie învăţătoare (Govora, 1642), ed. by Alin-Mihai Gherman, Bucharest, Editura Academiei Române, 2011. Floarea darurilor, in I. Gheţie, Al. Mareş (coord.), Cele mai vechi cărţi populare în literatura română, I, text stabilit de Alexandra Moraru şi Magdalena Georgescu, Bucharest, Editura Minerva, 1996.

On the Ethical Dative in Old and Modern Romanian

187

Gheţie, Ion (coord.), Texte româneşti din secolul al XVI-lea, Bucureşti, Editura Academiei Române, 1982. Istoria Ţării Româneşti de la octombrie 1688 până la martie 1717 (Cronica Anonimă despre Brâncoveanu), in Cronicari munteni, II, ed. by M, Gregorian, București, Editura pentru Literatură, 1961. Istoria Ţării Româneşti de stolnicul Constantin Cantacuzino, in Cronicari munteni, ed. by M. Gregorian, Bucharest, Editura pentru Literatură, 1961. Liturghierul lui Coresi (1570), ed. by Al. Mareş, Bucharest, Editura Academiei Române, 1969. Manuscrisul de la Ieud, ed. by Mirela Teodorescu, Ion Gheţie, Bucharest, Editura Academiei R.S.R., 1977. Neculce, Ion, Letopiseţul Ţării Moldovei, ed. by Iorgu Iordan, Bucharest, Editura de Stat pentru Literatură şi Artă, 1959. Palia de la Orăştie (1581-1582), ed. by Viorica Pamfil, Bucharest, Editura Academiei Române, 1968. Pravila ritorului Lucaci (1581), ed. by I. Rizescu, Bucharest, Editura Academiei Române, 1971. Psaltirea Hurmuzaki, vol. I, ed. by Ion Gheţie, Mirela Teodorescu, Bucharest, Editura Academiei Române, 2005. Sindipa, în I. Gheţie, Al. Mareş (coord.), Cele mai vechi cărţi populare în literatura română, I, text stabilit de Alexandra Moraru şi Magdalena Georgescu, Bucharest, Editura Minerva, 1996. Ureche, Grigore, Letopiseţul Ţării Moldovei, ed. by Mircea Scarlat, Bucureşti, Editura Minerva, 1978. Varlaam, Răspunsul împotriva Catihismusului Calvinesc (1645), in Opere, ed. by Mirela Teodorescu, Bucharest, Editura Minerva, 1984. Varlaam şi Ioasaf, in Cărţile populare în literatura românească, ed. by Ion Chiţimia, Dan Simionescu, Bucharest, Editura pentru Literatură, 1963.

References Albizu, P., 1997, “Generalized Person-case Constraint: A case for a syntax-driven inflectional morphology”, in: M. Uribe-Extebarria, A. Mendikoetxea (eds.) Theoretical issues on the morphology-syntax interface, ed. Donostia: Gipuzkoako Foru Aldundia/EHU, 1–33. Cardinaletti, A., M Starke, 1994, “The typology of structural deficiency. On the three grammatical classes”, University of Venice Working Papers in Linguistics, 4, 2, 41–109. Cuervo, M. C., 2003, Datives at large. Doctoral dissertation, Cambridge, Mass., MIT. Delbecque, N., B. Lamiroy, 1996, “Towards a typology of the Spanish dative” in: W. Van Belle, W. Van Langendonck (eds), Case and grammatical relations across languages: The dative, Amsterdam, John Benjamins, 73–117. Diaconescu, C. R., 2004, Romanian Applicative Constructions, Doctoral dissertation, Ottawa University.

188

Chapter Seven

Franco, J., S. Huidobro, 2008, “Ethical datives, clitic doubling and the theory of pro” in: J. Bruhn de Garavito, E. Valenzuela (eds), Selected Proceedings of the 10th Hispanic Linguistics Symposium, Sommerville, MA: Cascadilla Proceedings Project, 215–224. GALR = V. Guţu Romalo (ed.), 2005, Gramatica limbii române, vol. II, Bucharest, Editura Academiei Române. Jaeggli, O., 1982, Topics in Romance Syntax, Foris, Dordrecht. Jouitteau, M., M. Rezac., 2008, “The French ethical dative. 13 syntactic tests”, Bucharest Working Papers in Linguistics, IX, 1, 97–108. Niculescu, D., 2008, “Romanian Possessive Dative. The limits of the structure”, Revue roumaine de linguistique, LIII, 4, 485–515. Pylkkänen, L., 2002, Introducing Arguments, Doctoral dissertation, MIT. Rivero, M., 2003, “Reflexive clitic constructions with datives: syntax and semantics, Formal Approaches to Slavuc Linguistics (FASL) 11. The Amherst Meeting 2002, Michigan Slavic Publications, 469–494. Roberge, Y., M. Troberg., 2009, “The high applicative syntax of the dativus commodi / incommodi in Romance”, Probus, 21, 249–289. RGR = C. Dobrovie-Sorin, I. Giurgea (eds), 2012, A Reference Grammar of Romanian. Volume 1: The noun phrase, Amsterdam/Philadelphia, John Benjamins.

PART III: COMPLEMENTISERS, RELATIVE CLAUSES, AND COORDINATION

CHAPTER EIGHT TOWARDS A MICROPARAMETER C-HIERARCHY IN ITALO-ROMANCE VALENTINA COLASANTI1 St John’s College, University of Cambridge [email protected]

1. Introduction Complementation in Italo-Romance continues to represent an extremely interesting topic among Romance and theoretical linguists. Recently, much research has been directed towards the investigation of the fine structure of the left periphery on the basis of the rich dialectal variation offered by the Italian peninsula (Rizzi 1997; see also Benincà and Poletto 2004). Southern Italian dialects (SIDs), which make use of dual complementiser systems, make a significant contribution to our understanding of the discourse-domain from a synchronic and a diachronic perspective.2 In this respect the dialects of southern Lazio (SLDs) have not yet been considered. The syntactic microvariation exhibited in these upper-southern Italian dialects (USIDs) has generally gone unnoticed (Colasanti 2015). However, different varieties from Southern Lazio still display two distinct finite complementisers as shown in (1), (2) and (3). In 1

This work is funded by the St John’s College Cambridge Graduate Scholars grant (October 2015-September 2018). I would like to thank Adam Ledgeway for his extremely helpful comments and corrections on an earlier version of this paper. I am also grateful to the audiences of the Romance Syntax Workshop (University of Bucharest, November 2015), and the Romance Linguistics Seminar (Trinity Hall, Cambridge, January 2016), where earlier versions of this paper were presented. Any errors are the responsibility of the author. 2 Among others see Ledgeway (2003, 2005, 2012a, 2016); Benincà and Poletto (2004); Paoli (2004, 2007); Vecchio (2006); Damonte (2006, 2009); D’Alessandro and Di Felice (2015); D’Alessandro and Ledgeway (2010); Colasanti (2015).

192

Chapter Eight

what follows it will be argued that evidence from Southern Lazio proves essential for a proper understanding of complementation in ItaloRomance. Specifically, a microparameter C(omplementation)-hierarchy in Italo-Romance will be proposed. Before considering data from SLDs, however, it is important to review all the relevant facts concerning complementation across Italo-Romance that have already been investigated by previous scholars. (1)

a. b.

(2)

a. b.

(3)

a.

b.

m ao ditto ka ve addemane3 to-me they-have told that s/he-come tomorrow ͑ra meुुo ke li ͑rav͑mo fatto it-was better that them= we-were done Sonnino (M&S 2005)4 m avə diccə ka tu ve addumanə to-me they-have told that you come tomorrow ͑ mm͑ुुə kə mmə nə vavə it-is better that I= CL go.1SG Pontecorvo (M&S 2005) dikə ka Mariə vè5 I-say that Mario come.3SG.IND “I say that Mario comes” vogliə kə Maria vənissə I-want that Maria come.3SG.SUBJ “I want that Maria would come” Ceprano (Colasanti 2015)

1.1 Dual complementiser systems in Italo-Romance 1.1.1 Traditional descriptions Standard descriptions of Italo-Romance varieties highlight the presence of dual complementiser systems (Rohlfs 1969: 190; Tekavčić 1980: 446). In the dialects of the extreme South (henceforth ESIDs), namely Salento, central-southern Calabria and northern-eastern Sicily (province of Messina), finite complement clauses are introduced by one of two different complementisers according to the distribution of the dual 3

In what follows, free translations of the examples will be provided only where the sense cannot be immediately deduced from the glosses. 4 Manzini and Savoia (2005), henceforth M&S (2005). 5 Unless otherwise indicated, examples are from the author’s own fieldwork notes.

Towards a Microparameter C-Hierarchy in Italo-Romance

193

complementiser system found in Balkan varieties (i.e. Greek). In accordance with the Greek complementiser system, in these dialects we find realis declarative complements (traditionally marked with the indicative) introduced by ca ( -e after other adverbs, and ori < oare (Candrea and Densusianu 1914); later, Candrea (1931) explained oare and ori as the plural of oară: “sometimes ... sometimes..” > “either ... or”. Even if this solves the potential phonetic problems, the semantic development remains problematic, in my opinion. (ii) vare < (Proto-)Alb. vallë, a disjunctive particle similar to oare7 (Meyer 1891, Densusianu 1901; DER and DEX, s.v. oare, add, to the Latin etymon, “cf. alb. vallë”); but I could not find any clear attestation for the disjunctive and free-choice uses in Albanian (Rosetti 1968 cites a form vall “or”, but Çabej 1976:266, in the article on vallë, does not mention such a form); moreover, a different origin for oare and vare is not necessary, given that the alternation in oare/vare is found elsewhere (see above: veri/ori, voi/oi, vei/ăi, etc.). The free-choice marker cannot come directly from the verb “want”, because the morpheme order is FC-Wh, and not Wh-FC as expected if FC had originally been a verb (we can say “who wants, who you want”, but not “wants who, you want who”): compare Rom. ori-cine, oare-care “oriwho, oare-which” with Lat. qui-uis “who-want.2SG”, qui-libet “whopleases”, Sp. cual-quiera “which-wants(SUBJ)”, Cat. qual-se-vol “whoREFL/IMPERS-wants”, Alb. kush-do “who-want.2/3SG”. As shown by Dinică (2012), the order FC-Wh is expected if FC was originally a (correlative) disjunction. Note that the same coexistence of a FC-Wh order and a correlative disjunction use is found in Hungarian: (57)

a. b.

akár-ki “any(one), whoever” FC-who akár...akár “either...or”

Moreover, the element akár comes from the verb akar “want”, exactly like Rom. oare and ori (EWU derives akár from an imperative form of akar, translated in German as “wolle, wolle es!”). The various free-choice items of other Romance languages built with “want” all have the verbal part at the end, except OSp. sivuelqual, -que, -quando (see Lombard (1938), Dinică (2012), REW 9180 (for other 7 It expresses doubt, according to Buchholz and Fiedler (1987). They translate a vallë- question using German ob, which is, as we have seen in §1, a possible equivalent of oare- questions in which the hearer is not supposed to know the answer for sure: (i) Vallë vjen sot ai? ”Ob er wohl heute kommt?” (p. 395)

The Romanian Interrogative Particle oare

423

languages in which the “want”-element occurs at the end, see Haspelmath 1997). The OSp. forms sivuel-+Wh (attested only in Barceo—a 13th century author) are explained by Corominas and Pascual (1985) as reversed forms corresponding to Cat. Wh-Impers.se-wants—qualsevol, with metaphony sevuel>sivuel. I find this account unconvincing. Like in Rom. and Hung., we may start from an unattested disjunction sivuel < si/sic + uole(t). A disjunction derived from “want” is indeed attested in medieval Sp., with the novel verb “want” querer (< Lat. quaerere): quier, considered by Corominas and Pascual a calque (loan translation) of Lat. uel, without sufficient reasons—for the evolution ‘want’ > disjunction, see ORom. oare/vare and ori/veri, where a Latin loan translation is excluded; cf. also, composed with si, the form siquiera “at least; (not) even”. Sivuel may contain si “if; whether”—a conditional marker is expected in a disjunction specialized for relating conditional clauses (on this specialization for oare/vare, see the discussion below); moreover, a similar compound is attested in ORom.: săva < să ‘if; SUBJ’ + va ‘wants’ (see (58)-(59) for conditional+disjunction and disjunction uses, (60) for the free choice use; moreover, like Sp. siquiera, it also developed the meaning “at least, not even”): (58)

(59)

(60)

săva dobitoc fie, săva om, nu viiadze săva animal be.SUBJ.3 săva man not live.SUBJ.3 “Whether it is an animal or a human, it should not live” (Palia de la Orăştie [1581], Exodus 19.13) Să fie iară mielul fără beadă, SUBJ be.3 but lamb-the without flaw bărbătuş miel de un an, săva luaţi un ied. male lamb of a year or take.2PL a kid “The lamb should be with no flaw, a one year old male lamb, or take a kid.” (ibid., Ex. 12.5) săva ce veţi zice mie, FC what will.2PL tell me.DAT aceaia o voiu da that ACC.CL will.1SG give “Whatever you will ask for, I shall give you” (ibid. Genesis 34.11)

The evolution “want” > disjunctive particle is also attested in Lat. uel, which is explained as an old second person indicative of uelle “want”: *uels(i) > *uell > uel (Leumann 1977:142; the classical form uīs would

424

Chapter Fourteen

have been taken from another verb, ibid.:526) or a 2sg imperative of uelle (Kühner and Stegmann 1912-1914, II, 2, 107). Note that a 2nd imperative origin is also possible for Ro.: *uole (the imperative of Vulg. Lat. *uolēre) > *voare, identical with the result of the 3rdsg. *uolet after the fall of -t (this hypothesis has not been considered by Romanian historical linguists, presumably because want lacks an imperative in MRom., classical Lat., as well as other languages). Oare is indeed attested as a multiple (“correlative”) disjunction in ORom., similar to Lat. uel. Moreover, disjunctive oare/vare predominantly appears inside restrictions of operators based on implication, i.e. in environments where disjunction is equivalent with the conjunction of the alternatives: (61) (62)

(63)

(64)

(A › B) → C |= (A → C) š (B → C) oare să veţi vrea, oare să nu veţi oare if will.2PL want oare if not will.2PL vrea, voi să faceţi aceasta want you.PL SUBJ do.2PL this’ “Whether you want it or not, you should do this” (Coresi, Evanghelie cu învăţătură [1581], 67) Cine va vrea – ori bărbat ori muiare, who will want ori man ori woman oare bogat oare sărac, ori mai mare ori mai mic oare rich oare poor ori more big ori more little – pre acea cale bună, să vie on that way good SUBJ come.3 după mine! after me “Who will want—man or woman, rich or poor, bigger or smaller—to (go) that good way, should follow me!” (ibid., 67) Tot omul, când înceape oare cuvânt, every human-the when starts oare word oare lucru, dentâiu are grije şi scârbă, oare work first has worry and distress iară la săvârşitul lucrului bucură-se but at end-the work-the.GEN rejoices-REFL şi se veseleaşte (ibid., 77) and REFL is-cheerful “Any man, when he starts a word or a work, is first worried and distressed, but when the work is done he rejoices and is cheerful”

The Romanian Interrogative Particle oare

(65)

(66)

425

vă rugăm toţi sfenţi părinţi you.PL.ACC pray.1PL all holy parents oare vlădici, oare episcopi, oare popi oare rulers oare bishops oare priests în cărora mână va veni aceastea cărţi creştineşti in whose hand will arrive these books Christian cum mainte să cetească, necetind that first SUBJ read.3 not-reading să nu judece SUBJ not judge.3 “We ask you all, holy parents, or rulers, or bishops, or priests, into whose hands these Christian books will fall, to read first, and not judge without having read.” (Coresi, Tetraevanghel [1561], 249r) staţi şi ţineţi învăţăturile stay.2PL and keep.2PL teachings-the cu carele v-aţi învăţat, vare în cuvânt, with which REFL-have.2PL tought vare in word vare cu cărţile noastre. vare with letters-the our “Stand fast and keep the teachings you have been taught, whether by word or by our letters.” (Noul Testament de la Bălgrad [1648], 278r)

It also appears in other contexts (see (67)), but the restriction environment predominates—in my 16th-century corpus, I found 17 examples in restriction contexts and only 3 in other environments. (67)

Şi atunce oare se va proslăvi, and then oare REFL will.3SG praise oare se va ruşina oare REFL will.3SG disgrace “And then he will be either praised, or disgraced.” (Coresi, Evanghelia cu învăţătură [1581], 113)

In the 1648 New Testament I found disjunctive vare only inside restrictions (the form oare no longer appeared as a disjunction). Sometimes oare/vare suffices to express the conditional:

426

(68)

Chapter Fourteen



iubeşti tu pre alt om, love.2SG you ACC another man vare ţi-e tată, vare mumă, vare fraţi, vare you.DAT-is father vare mother vare brothers vare surori, vare feciori, vare feate, vare vare sisters vare sons vare daughters vare cumnaţi, macară în ce chip de rudă, brothers-in-law any in what sort of relative sămânţă au striin, de o leage vare de altă seed or alien of one law vare of another leage cu tine, priiatnic, frăţie, law with you friend comrade miluitoriul tău au vrăjmaşul tău, creştin vare benefactor-the your or enemy-the your Christian vare păgân (Coresi, Tâlcul Evangheliilor [1567-1568], 133v) pagan “You should love another person, whether it is your father, or your mother, or brother, or sister, or son, or daughter, or brother in law, or any sort of relative, kinsman or alien, of the same religion as you or of a different one, friend, comrade, your benefactor or your enemy, Christian or pagan” SUBJ

This environment might explain how verbal forms of “want” became disjunction markers: “he/you may want X, he/you may want Y, in both cases Z” → “whether X or Y, then Z”. Indeed, Lat. uel is also often found in this environment: (69)

ne nimio opere sumat lest too-much.ABL effort.ABL take.SUBJ.3SG operam si quem conuentum work.ACC if somebody.ACC met.ACC uelit uel uitiosum uel sine uitio wants uel vicious.ACC or without vice uel probum uel improbum uel honest.ACC uel dishonest.ACC “(..) lest he should pay too much effort if he wants to meet somebody, a vicious person or one with no vice, a honest person or a dishonest one” (Plautus, Curculio, 468-469)

This use explains the development of the concessive meaning “even” for uel (also found in ORom. săva).

The Romanian Interrogative Particle oare

427

Moreover, this “inclusive” environment (in the sense that something holds for all alternatives) underlies the evolution into free choice markers. ORom. oare/vare had recently been grammaticalized as a free choice marker, as shown by the fact that it still could appear separated from the pronoun by a preposition: (70)

a.

b.

cum să strice oamenilor so-that SUBJ harm.3 people-the.DAT oare în ce chip poate oare in what way can.3SG “in order to harm people in any way he can” (Coresi, Tâlcul Evangheliilor [1567-1568], 30v) Vare în ce casă veţi întra, vare in what house will.2PL enter întâiu ziceţi: pace casei first say.IMPV.2PL peace house-the.DAT aceştiia this “In whatever house you will enter, first say: Peace to this house!” (Noul Testament de la Bălgrad [1648], 81v)

Modern Romanian uses another form with a verbal origin as a correlative disjunction, in the contexts in (62)-(68): fie (= the 3rd person subjunctive of fi ‘be’). Interestingly, this form too yielded a preposed freechoice marker, further developed into a universal marker:8 (71)

fie-care be.3.SUBJ-which “each, every(one)”

Concerning now the evolution oare/vare, in the 16th century the interrogative use is already attested, but rare. In the Old Romanian corpus I used, I found 7 examples of the interrogative use, always sentence-initial (at most preceded by a subordinate clause) and in polar questions:

8

Admittedly, disjunctive fie is attested later (1646) than universal fie- (16th century), but we may assume that a disjunctive fie existed, at some point, in the language, during the long period for which there are no attestations (between late Antiquity and the 16th century).

428

(72)

Chapter Fourteen

Oare afla-va fapte bune pre pământ? oare find-will.3SG deeds good on Earth ‘Will he find good deeds on Earth?’ (Coresi, Tâlcul Evangheliilor [1567-1568], 242r)

The current interrogative particles at this stage were au (see §3 below) and doară.9 In the second half of the 17th century, the use of interrogative oare increases. It is still predominantly sentence-initial and used in polar questions, but other uses start to appear. In an extensive fragment of the 1688 Bible (the “Bucharest Bible”), I found 26 examples of clause-initial oare in polar questions, and 12 examples in wh-questions, in which oare precedes the wh-word in 9 cases and follows it in 3 cases. Moreover, in 2 examples it is used to introduce an indirect interrogative, probably as an instance of free direct speech. In the same text, the indefinites in oare- are currently used only with an indifference/ ignorance existential reading, the universal free choice reading (attested in the 16th century, see (70) above) being lost. The variant vare has completely disappeared. The existential reading of oare- indefinites is attested, for some forms, as early as the 16th century (Dinică 2012): (73)

Şi unul oarecine, un tânăr, mearge după el and one oare-who a young walks behind him îmbrăcat întru o cămaşe dressed in a shirt “And some young man is walking behind him dressed in a shirt.” (Coresi, Tetraevanghel [1561], 104r)

Corpus George Bacovia, Zborul cărţilor [1915]. In Opere, Bucharest, Editura Minerva, (Scriitori români), 1978, 238–240. Biblia ortodoxă = Biblia sau Sfânta Scriptură. Bucharest, Editura Institutului Biblic şi de Misiune al Bisericii Ortodoxe Române, 1994. Mateiu Ion Caragiale, Craii de Curtea-Veche [1929]. In Opere. Bucharest, Editura Fundaţiei Culturale Române, 1994, 55–162. G. Călinescu, Bietul Ioanide [1953]. In Opere, I: vol. 5, 5–423, II: vol. 6, 5–427. Bucharest, Editura pentru Literatură, 1967.

9

For a comparison of the frequencies of oare and other interrogative particles during the evolution of Old Romanian, see Gheorghe (2016).

The Romanian Interrogative Particle oare

429

G. Călinescu, Enigma Otiliei. [1938] I: In Opere, vol. 3. Bucharest, Editura pentru Literatură, 1966, p. 5–306; II: ibid., vol. 4. Bucharest, Editura pentru Literatură, 1966, p. 5–291. Constantin Chiriţă, Cireşarii, I. Cavalerii Florii de Cireş. Bucharest, Albatros, 1972. Coresi, Evanghelia cu învăţătură [1581], ed. by Sextil Puşcariu, Alexie Procopovici, Bucharest, 1914. Coresi, Tâlcul evangheliilor. Molitvenic rumânesc [1567–1568]. Ediţie critică de Vladimir Drimba. Bucureşti, Editura Academiei Române, 1998, p. 31–187; ibid. p. 189–211. Coresi, Tetraevanghel [1561], ed. by Florica Dimitrescu, Bucharest, Editura Academiei, 1963. Petru Cimpoeşu Povestea Marelui Brigand. Cluj-Napoca, Editura Dacia, 2000. Dan Doboş, Abaţia. Bucharest, Editura Nemira, 2002. Flavius Guiaş, A fi sau a nu fi... modern, Antimis, Editura Virtuală. Calistrat Hogaş, Pe drumuri de munte. [1912] Bucharest, Fundaţia Regală pentru Literatură şi Artă, 1944 Interacţiunea verbală în limba română actuală. Corpus (selectiv). Schiţă de tipologie, coord. by L. Ionescu-Ruxăndoiu. Bucharest: Editura Universităţii din Bucureşti, 2002. Noul Testament, Bălgrad [1648]. Alba Iulia, Editura Episcopiei Ortodoxe Române a Alba Iuliei, 1988. Palia, Orăştie [1582]. Ed. by Viorica Pamfil. Bucharest, Editura Academiei, 1968. Petre Ispirescu, Basme, snoave şi glume [1883]. In Opere, I, 425–453. Bucharest, Editura Minerva, 1971. Gib Mihăescu, Donna Alba [1935]. In Opere, vol. 4. Bucharest, Editura Minerva, 1980, p. 393–791. Ion Minulescu, Roşu, galben şi albastru [1924]. Bucharest, Editura Minerva, 1991. Hortensia Papadat-Bengescu, Fecioarele despletite [1925]. In Opere. III, Bucharest, Editura Minerva, 1979, 7–155. Camil Petrescu, Ultima noapte de dragoste, întâia noapte de război [1930]. In Opere. II. Bucharest, Editura Minerva, 9–328. Liviu Rebreanu, Pădurea spânzuraţilor [1922]. In Opere, vol. 5. Bucharest, Editura Minerva, 1975. Liviu Rebreanu, Răscoala [1932]. In Opere, vol. 8. Bucharest, Editura Minerva, 1975. Ionel Teodoreanu, La Medeleni [1925-1927]. In Opere, vol. 1-3. Bucharest, Editura Minerva, 1968.

References Barbu, A., 1999, “Complexul verbal”, Studii şi cercetări lingvistice, 50, 1, 39–84. Beijering, K., 2012, Expressions of epistemic modality in Mainland Scandinavian, PhD Dissertation, University of Groningen. Bianchi, V., S. Cruschina, 2016, “The derivation and interpretation of polar questions with a fronted focus”, Lingua, 170, 47–68.

430

Chapter Fourteen

Çabej, E., 1976, Studime gjuhësore në fushë të shqipes, II: P-ZH, Prishtinë, Rilindja. Candrea, I.-A., O. Densusianu, 1914, Dicţionarul etimologic al limbii române. Elementele latine (A–Putea), Bucharest, Socec. Candrea, I.-A., 1931, Dicţionarul enciclopedic ilustrat. I: Dicţionarul limbii române din trecut şi de astăzi, Bucharest, Cartea Românească. Ciorănescu, A., 1966/2001, Diccionario etimológico rumano, La Laguna, Tenerife, Biblioteca Filológica, Universitad de la Laguna, 1966. [Romanian translation by Tudora Şandru Mehedinţi and Magdalena Popescu Marin, Bucureşti, Editura Saeculum, 2001.] Corominas, J., J. A. Pascual, 1985, Diccionario crítico etimológico castellano e hispánico. IV: ME-RE, Madrid, Editorial Gredos (first edition: 1981). DA (1913) = Dicţionarul limbii române, coord. by S. Puşcariu, vol. I, A-B, Bucharest: Librăriile Socec & Comp. şi C. Sfetea. Densusianu, O., 1901, Histoire de la langue roumaine, I. Les Origines, Paris, Ernest Leroux. DEX = Dicţionarul explicativ al limbii române, coord. by I. Coteanu, L. Seche, M. Seche, Bucharest, Editura Academiei, 1975. Dinică, A., 2012, Pronumele şi adverbele nehotărâte în limba română, PhD Dissertation, University of Bucharest. DLR = Academia Română, Dicţionarul limbii române. Volume VII, second part: letter O, Bucharest, Editura Academiei, 1969. Dobrovie-Sorin, C., 1994. The Syntax of Romanian, Berlin, Mouton de Gruyter. Embick, D., R. Noyer, 2001, “Movement Operations after Syntax”, Linguistic Inquiry, 32, 555–595. Ernst, T., 2004, The Syntax of Adjuncts, Cambridge, Cambridge University Press. Erteschik-Shir, N., 2010, “The phonology of adverb placement, object shift, and V2: The case of Danish ‘MON’”, in: N. Erteschik-Shir, L. Rochman (eds), The Sound Patterns of Syntax, Oxford, Oxford University Press. EWU = Etymologisches Wörterbuch des Ungarischen, ed. by Benkő Loránd, vol. 1–2, Budapest, Akadémiai Kiadó, 1993–1995. GALR = Academia Română, Institutul de Lingvistică „Iorgu Iordan—Al. Rosetti”, Gramatica limbii române, Valeria Guţu Romalo (ed.), Bucharest, Editura Academiei Române, 2008. Gheorghe, M., 2016, “Polar and wh-interrogatives. Indirect interrogatives. Exclamatory constructions”, in: G. Pană Dindelegan (ed.), The Syntax of Old Romanian, Oxford, Oxford University Press, 576–587. Giurgea, I., 2011, “The Romanian verbal cluster and the theory of head movement”, in: J. Herschensohn (ed.), Romance Linguistics 2010. Selected papers from the 40th Linguistic Symposium on Romance Languages (LSRL), Seattle, Washington, March 2010, Amsterdam, Philadelphia, John Benjamins, 271–286. Giurgea, I., 2017, “Preverbal Subjects and Topic Marking in Romanian”, Revue roumaine de linguistique, LXII, 3, 279–322.

The Romanian Interrogative Particle oare

431

Giurgea, I., E.-M. Remberger, 2016, “Illocutionary Force”, in: A. Ledgeway, M. Maiden (eds), The Oxford Guide to the Romance Languages, Oxford, Oxford University Press, 863–878. GLR (2013) = G. Pană Dindelegan (ed.), The Grammar of Romanian, Oxford, Oxford University Press. Haspelmath, M., 1997, Indefinite Pronouns, Oxford, Clarendon Press. Hill, V., 2003, “Discourse markers in interrogative clauses”, Balkanistica, 16, 71– 96. Kühner, R., C. Stegmann, 1912–1914, Ausführliche Grammatik der lateinischen Sprache, Hannover/Leipzig, second edition. Leumann, M., 1977, Lateinische Laut- und Formenlehre, new edition (first edition: 1926-1928), München, C.H. Beck. Lombard, A., 1938, “Une classe spéciale des termes indéfinis dans les langues romanes”, Studia Neophilologica, 11, 1–2, 186–209. Meyer, G., 1891, Etymologisches Wörterbuch der albanesischen Sprache, Straßburg, Karl Trübner. Ordbok over det danske sprog, available online at http://ordnet.dk/ods/ordbog Panaitescu, M., 2013, The Syntax and Semantics of Free-Choice Items in Romanian: A Comparison with English, PhD Dissertation, University of Bucharest. Philippide, A., 1927, Originea românilor, II, Iaşi, Viaţa Românească. Pierrehumbert, J. B., 1980, The phonetics and phonology of English intonation, PhD Dissertation, MIT. REW = W. Meyer-Lübke, Romanisches Etymologisches Wörterbuch, Heidelberg, Carl Winters, 1936. Rivero, M. L., 2001, “Last Resort and V Movement in Balkan Languages”, in: M. L. Rivero, A. Ralli (eds), Comparative Syntax of Balkan Languages, Oxford, Oxford University Press, 200–223. Rizzi, L., 2001, “On the position “Int(errogative)” in the left periphery of the clause”, in: G. Cinque, G. Salvi (eds), Current Studies in Italian Syntax. Essays offered to Lorenzo Renzi, Amsterdam, Elsevier, 287–296. Rosetti, A., 1968, Istoria limbii române. De la origini până în secolul al XVII-lea, Bucharest, Editura pentru Literatură. Scriban, A., 1939, Dicţionaru limbii româneşti, Iaşi, Institutul de Arte Grafice „Presa Bună”. Tiktin, H., 1903, Rumänisch-Deutsches Wörterbuch, Bucharest, Imprimeria Statului. Truckenbrodt, H., 2006, “On the semantic motivation of syntactic verb movement to C in German”, Theoretical Linguistics, 32, 257–306. Vasilescu, A., I. Vântu, 2008, “Tipuri de enunţuri în funcţie de scopul comunicării”, in: GALR, II, 25–46. Zafiu, R., 2008, “Modalizarea”, in: GALR, II, 702–726. Zafiu, R., 2013, “Modality and evidentiality”, in: GLR, 575–584. Zamfir, D. M., 2014, “Concordanţe româno-albaneze în gramaticalizarea verbului a vrea”, Conference given at the Romanian Academy.

CHAPTER FIFTEEN ON THE MARKING OF NEGATIVE PRESUPPOSITION IN REGIONAL VARIETIES OF ITALIANx FEDERICA COGNOLA Sapienza University, Rome [email protected]

NORMA SCHIFANO University of Cambridge [email protected]

1. Introduction Manner adverbs are known to have developed discourse values both in Romance (Belletti 1990; Lonzi 1991; Cinque 1991, 1999; Coniglio 2008; Cardinaletti 2011; Waltereit and Detges; Hernanz 2010; Padovan and Penello 2014, a.o.) and Germanic (Weydt 1969; BaardewykResseguier 1991), and Italian (henceforth It.) ben is no exception. By way of illustration, consider the examples below, showing that the interpretation conveyed by ben (1) differs from the prototypical manner reading of its full counterpart bene ‘well’ (2):1 x

This paper stems from joint research. For academic concerns, Norma Schifano takes responsibility for sections 1 and 2 and Federica Cognola for sections 3 and 4. 1 The translations provided in (1) are those offered in the cited works. The differences among distinct authors underlie the difficulty of rendering the meaning of ben in English, where the closest approximation could be an emphatic stress on the verb or the use of indeed, e.g. Avrei ben aiutato Maria, se avessi saputo ‘I WOULD have helped / I would indeed have helped Maria, if I had known’. Because of the lack of exact correspondence, ben will not be translated in the

434

(1)

(2)

Chapter Fifteen

a.

Dagli ben una risposta! (Zanuttini 1997:135) “Do give him an answer!” b. Gianni avrà ben risposto (Belletti 1994:30) “Gianni will have indeed answered” c. Deve ben aver già confessato (Coniglio 2008:123) “He MUST have already confessed” Silvia non cucina bene S. not cooks well “Silvia does not cook well”

In the literature, the interpretation expressed by It. ben and its Romance cognates has often been linked to the notion of emphatic assertive meaning, whereby ben has been analysed as an element reinforcing the assertion (Belletti 1990, 1994) and expressing the speaker’s confidence about the propositional content of their assertion (Coniglio 2008; Cardinaletti 2011), thus operating on the positive polarity of the sentence (Battlori and Hernanz 2013 on Spanish bien). Conversely, in Cognola and Schifano (forthcoming) we have carried out an investigation on the distribution and interpretation of ben in Trentino regional Italian (henceforth Tr.) when used as a discourse particle and we have shown that its core property is that of denying the interlocutor’s negative presupposition (cf. also Waltereit and Detges 2007 on French and Hernanz 2010 on Spanish), i.e. Tr. ben can only occur in (syntactically positive) contexts in which the negative counterpart of the proposition expressed by the sentence is part of the common ground, as shown by the contrast in (3) and (4) (adapted from Cognola and Schifano forthcoming): (3)

a.

Speaker A:

b.

Speaker B:

Battiston arriverà, Battiston will.arrive secondo me according to.me “Battiston will arrive, I think” (positive presupposition) #Arriverà ben he.will.arrive BEN “He will arrive”

examples below. Translations will be provided only when required by the discussion.

On the Marking of Negative Presupposition

(4)

a.

Speaker A:

b.

Speaker B:

435

arriverà Battiston, not will.arrive Battiston secondo me according to.me “Battiston will not arrive, I think” (negative presupposition) Arriverà ben he.will.arrive BEN “He will arrive”

In the present study, we aim to discuss the results of a parallel investigation that we carried out across further localities of the Italian peninsula in order to broaden our understanding of the intepretation and syntactic distribution of this element in regional Italian and to contribute to the wealth of literature on this topic, which still awaits an in-depth investigation into Italo-Romance. In line with Cognola and Schifano (forthcoming), we take ben in contexts like (1)-(4) and those tested below to be a discourse particle, i.e. an element which conveys information about the epistemic state of discourse participants and does not contribute to the descriptive content of the utterance (Zimmerman 2011). The discussion is organised as follows. Section 2 offers an overview of the results gathered with a questionnaire run with native speakers of different varieties of regional Italian. Focusing our attention on Group 3, i.e. the set of varieties under examination in the present paper, we describe a number of restrictions affecting the distribution of ben across distinct T(ense)A(spect)M(ood) contexts of occurrence, as well as its syntactic placement. Section 3 deals instead with the semantic/pragmatic interpretation of this element. As ben of Group 3 denies a negative presupposition and is primarily connected with negation, we extend the same analysis suggested for Tr. ben in Cognola and Schifano (forthcoming), i.e. ben is located in NegPresuppositionalP and is licensed by a negative operator in ForceP. Nevertheless, the two groups exhibit a crucial interpretative distinction regarding the nature of the denied presupposition. Building on a comparison with German doch, we link the restrictions which affect ben in Group 3 but not in Trentino to this interpretative property. Section 4 offers a summary of the main findings of the paper.

436

Chapter Fifteen

2. The syntactic distribution In order to assess the syntactic distribution of ben in regional Italian, we have carried out a comprehensive investigation with 28 native speakers of mixed age, gender and education from 15 different localities (see Table 1.1), who were asked grammaticality judgments on a 1-to-5 scale (1: fully grammatical; 5: fully ungrammatical) on 67 sentences testing the occurrence of ben across a variety of verb forms and TAM contexts. The tested verb forms include unaccusative and transitive lexical vs auxiliary verbs, as well as restructuring ones, while the TAM contexts include the indicative (present, imperfect, simple future, future perfect, present perfect, pluperfect), the conditional (present, perfect), the subjunctive (present, pluperfect), as well as the present imperative. The two tested unaccusative verbs, cf. arrossire ‘to blush’ (henceforth unaccusative A) and arrivare ‘to arrive’ (henceforth unaccusative B), are a scalar and a non-gradable verb, respectively, chosen to check whether Hernanz’s (2010:42, fn.29) remark on Catalan ben, which is often allowed with verbs denoting scalar properties but not with non-gradable predicates, applies to Italian as well. As for the tested restructuring verbs (cf. volere ‘want’, potere ‘can’, smettere ‘stop’), these lexicalise distinct positions in Cinque’s (2006:93) hierarchy of restructuring, as reported in (5) (adapted from Ledgeway forthcoming) and were chosen to test whether their relative position within the IP clausal spine affects the occurrence of ben: (5)

[ModEpistemic/Alethic must/can [AspHabitual be wont [AspPredispositional tend [AspRepetitive do again [ModVolition want [AspTerminative stop [AspContinuative(I) continue [AspDurative/Progressive stand [ModObligation/Ability must/can [AspFrustrative/Success succeed [ModPermission can [AspConative try [Causative do [AspInceptive begin [AspAndative go [AspCompletive finish [v-VP V…

In all contexts and with all verb forms, judgments have been asked about the placement of ben both in pre and post-verbal positions, as well as before and after the non-finite form selected by functional verbs. Whenever possible, the same verb form has also been tested with different intepretations (e.g. future indicative with temporal vs epistemic intepretation).

On the Marking of Negative Presupposition

437

Table 1.1 Geographic distribution of informants Region

Province

Trentino-Alto Adige

Trento

Veneto

Belluno

Venice Verona

Friuli-Venezia Giulia Lombardia Piemonte Emilia Romagna Marche Lazio Puglia

Vicenza Rovigo Pordenone Udine Milano Bergamo Asti Ravenna Ancona Frosinone Bari

Locality and number speakers Trento (5) Cles (3) Civezzano (1) Rovereto (1) Agordo (1) Cortina d’Ampezzo (1) Feltre (1) Mestre (1) Verona (1) Peschiera del Garda (1) Villafranca di Verona (1) Valdagno (1) Rovigo (1) Pordenone (1) Udine (1) Milano (1) Nasolino (1) Asti (1) Ravenna (1) Sassoferrato (1) Frosinone (1) Bari (1)

of

In accordance with their judgments, speakers can be divided into three groups, i.e. Group 1 (speakers from Trentino), Group 2 (speakers from Rovereto, Agordo, Cortina D’Ampezzo, Feltre, Mestre, Peschiera del Garda, Verona, Valdagno, Villafranca di Verona and Pordenone) and Group 3 (speakers from Rovigo, Udine, Milano, Nasolino, Asti, Ravenna, Frosinone, Sassoferrato and Bari). Group 3 will be the object of the present study, while the reader is referred to Cognola and Schifano (2015, forthcoming) for a discussion of Group 1 and 2. The results of this investigation for Group 3 (1 speaker for each of the 9 varieties) are summarised in Table 1.2, where we observe that 4 and 5 were given only in 18.1% and 16.8% of the tested contexts, respectively:2 2

For the central-southern speakers belonging to Group 3, judgements refer to their compentence of standard Italian. Conversely, ben is not productive in their regional varieties of Italian nor in their local Romance dialects, in none of the tested contexts. It must also be noted that for these speakers the reduced form ben

438

Chapter Fifteen

Table 1.2 Overview of judgments in Group 3 (9 speakers / 67 sentences) Judgment Bari

1 46%

2 26%

3 27%

4 0%

5 0%

Asti Milano

0% 20%

5% 28%

28% 36%

36% 13%

0% 1%

Rovigo

56%

10%

17%

10%

4%

Frosinone

40%

1%

13%

33%

10%

Ravenna

1%

4%

49%

28%

15%

Udine

24%

15%

11%

18%

27%

Nasolino

15%

13%

15%

21%

33%

Sassoferrato

17%

8%

7%

4%

62%

Total

24.3%

12.2%

24.7%

18.1%

16.8%

The accepted contexts, namely sentences rated between 3 and 5 by at least 8 out of 9 speakers, are summarised in Table 1.3, while the corresponding sentences are reported in (6):3 Table 1.3 Summary of accepted contexts in Group 3 TAM context present perfect

Interpretation present perfect

conditional perfect

counterfactual

future perfect

temporal epistemic

potere ‘can’ present indicative potere ‘can’ imperfect indicative

present counterfactual

Verb form transitive (8/9) (6a) transitive (9/9) (6b) transitive (8/9) (6c) transitive (8/9) (6d) transitive (8/9) (6e) transitive (8/9) (6f)

would not be morpho-phonologically well-formed in their local repertoire. We thank Gigi Andriani and Valentina Colasanti (p.c.) for insightful comments on this point. 3 The numbers between brackets in the ‘verb form’ column indicate the number of speakers who accepted the tested context over the total number of interviewed speakers, e.g. 8/9 indicates that 8 out of 9 speakers accepted it.

On the Marking of Negative Presupposition potere ‘can’ simple future

temporal epistemic

potere ‘can’ conditional

conditional

potere ‘can’ conditional perfect potere ‘can’ future perfect volere ‘want’ simple future

counterfactual epistemic temporal epistemic

volere ‘want’ pluperfect indicative

(6)

a. b.

c.

d.

e.

f.

temporal

439

transitive (8/9) (6g) transitive (8/9) (6h) transitive (8/9) (6i) transitive (8/9) (6j) transitive (8/9) (6k) transitive (9/9) (6l) transitive (9/9) (6m) transitive (8/9) (6n)

Gianni ha ben comprato qualcosa G. has BEN bought something Gianni avrebbe ben comprato G. would.have BEN bought qualcosa, se avesse potuto something if he.had been.able Gianni avrà ben comprato G. will.have BEN bought qualcosa per quando noi torniamo something for when we come.back Gianni avrà ben comprato G. will.have BEN bought qualcosa, immagino something I.think Gianni può ben parlare con la G. can BEN speak with the povera Maria poor M. Gianni poteva ben parlare con G. could.IMP BEN speak with la povera Maria the poor M.

440

Chapter Fifteen

g.

h.

i.

j.

k.

l.

m.

n.

Gianni potrà ben parlare con G. will.be.able BEN speak with il direttore quando arriva in ufficio the director when he.arrives in office Gianni potrà ben parlare con G. will.be.able BEN speak with il direttore se ha un problema, the director if he.has a problem non devo sempre farlo io not I.must always do=it I Gianni potrebbe ben parlare con G. could.COND BEN speak with la povera Maria the poor M. Gianni avrebbe ben potuto G. could.have BEN been.able licenziare Maria, ma non lo fire M. but not it ha fatto has done Gianni avrà ben potuto G. will.have BEN been.able licenziare Maria immagino, visto che fire M. I.think given that è il capo he.is the boss Gianni vorrà ben incontrare G. will.want BEN meet Rihanna quando arriva in aeroporto R. when she.arrives in airport Gianni vorrà ben incontrare Rihanna, G. will.want BEN meet R. non è mica così timido not he.is not so shy da tirarsi indietro to back down Gianni aveva ben voluto G. had BEN wanted incontrare Rihanna meet R.

On the Marking of Negative Presupposition

441

This result sharply contrasts with the judgments expressed by the speakers of Group 1, who rated 90,3% of the tested contexts as fully grammatical. Accordingly, in Cognola and Schifano (forthcoming) we identified Trentino as the productive isogloss for this phenomenon. The interesting question amounts therefore to the identification of the restrictions affecting the occurrence of ben in Group 3. If we leave aside minor cases of expected individual variation, we first observe that ben is excluded in the only two non-root embedded clauses included in the questionnaire, featuring a present (7a) and pluperfect (7b) subjunctive, respectively: (7)

a.

b.

*Gianni vuole che G. wants that ben qualcosa BEN something *Gianni avrebbe G. would.have avesse ben had.SBJV BEN qualcosa per cena something for dinner

Marco compri M. buys.SBJV per cena for dinner voluto che Marco wanted that M. comprato bought

In addition to this excluded context (henceforth Restriction 1), which is the only one shared by all the three groups (see Cognola and Schifano forthcoming on Group 1), Group 3 exhibits three additional restrictions. First, Table 1.3 shows that the accepted contexts among non-restructuring verbs mainly include TAM combinations involving a non-finite form (henceforth Restriction 2), i.e. the participle or the infinitive selected by potere, with the exception of the pluperfect indicative, accepted only by half of the varieties included in this group.4 A selection of the tested sentences including a simple verb form is reported below, which were rated between 1 and 2 by all varieties of Group 3: (8)

4

a.

*Gianni G. cena dinner

compra buys quando when

qualcosa something può (Group 3) he.can

ben

BEN

per for

Restriction 2 applies to Group 2 as well (see figures in Cognola and Schifano 2015).

442

Chapter Fifteen

b.

c.

d.

e.

*Gianni comprava ben qualcosa G. bought.IPFV BEN something per cena quando poteva for dinner when he.could *Gianni comprerà ben qualcosa per G. will.buy BEN something for cena quando siamo via dinner when we.are away *Gianni comprerebbe ben qualcosa G. would.buy BEN something per cena, se potesse for dinner if he.could *Compra ben qualcosa per cena! buy.IMP.2SG BEN something for dinner

Second, among the tested restructuring verbs, volere ‘want’ proves to be more restricted than potere ‘can’, being allowed by (almost) all varieties only with the simple future and pluperfect indicative, although the other tenses still score quite high, being allowed by more of half of the varieties. In this case, it is not possible to observe a preference for simple tenses over compound ones, nor for modally marked contexts over unmarked ones. Conversely, smettere ‘stop’ is never accepted by any of the varieties under examination in Group 3, although compound tenses score better than simple tenses and modally marked intepretations score better than temporal ones (e.g. smettere with a future simple is accepted by 3 out of 9 varieties but the number increases to 5 out of 9 if the future conveys an epistemic meaning). We will refer cumulatively to this set of restrictions affecting restructuring verbs as Restriction 3. Finally, we observed that unaccusatives are less preferred than transitives (e.g. the present perfect is accepted by 8 out 9 speakers with transitives but only by 5 out of 9 speakers with unaccusatives) and unaccusative A (cf. arrossire ‘to blush’) is less preferred over unaccusative B (cf. arrivare ‘to arrive’) (e.g. the future perfect with epistemic meaning is accepted by 8 out of 9 speakers with transitive verbs, by 7 out of 9 speakers with unaccusative B and only by 3 out of 9 speakers with unaccusative A) (Restriction 4)5. A broad summary of these restrictions is offered in (9): (9) 5

Generalizations about distributional restrictions of ben in Group 3

Unaccusative are less preferred than transitives in Group 2 as well (see figures in Cognola and Schifano 2015). As for the scalar vs non-gradable distinction, our findings do not match Hernanz’s remark on Catalan ben (cf. §2).

On the Marking of Negative Presupposition

443

(i) embedded non-root contexts are ruled out (cf. Restriction 1); (ii) TAM combinations not involving a non-finite form are ruled out (cf. Restriction 2); (iii) among restructuring verbs, potere is widely accepted, volere is more restricted and smettere is largely ruled out (cf. Restriction 3); (iv) among non-restructuring verbs, unaccusatives are less preferred than transitives (cf. Restriction 4). As for its syntactic placement, ben is systematically excluded from the preverbal position across all the groups (see Table 1.4) and sits in a position comprised between the one targeted by the lexical, auxiliary and restructuring verbs and that hosting the non-finite form selected by functional verbs: (10)

a. b. c.

[IP (*ben) Vlex ben] [IP (*ben) Aux ben Prt] [IP (*ben) Vrestr ben Vnon-fin]

Table 1.4 Judgements on ben - Vlex / AUX / Vrestr across all groups (28 speakers / 67 sentences) Judgement

1

2

3

4

5

Total

1792

88

39

7

6

Total %

92.7%

4.5%

2%

0.3%

0.3%

To sum up, our parallel investigation into the use of ben across further varieties of regional Italian has shown that, outside the productive isogloss of Trentino (cf. Group 1), the use of ben as a discourse particle enjoys a more restricted distribution in terms of admitted verb forms and TAM contexts, while its syntactic placement remains consistent across all the groups.

3. The semantic/pragmatic interpretation While existing literature on It. ben and its Romance cognates has focused its attention on the emphatic affirmative value of such particles (see references in §1), in Cognola and Schifano (forthcoming) we have argued that Tr. ben is inherently presuppositional, i.e. by emphasising the truth of the proposition, it denies the speaker’s negative presupposition, as

444

Chapter Fifteen

exemplified in (3)-(4) above (see Hernanz 2010 on the same effect for Spanish bien). The interpretation of ben as used in Group 3 can be characterised in the same way, i.e. in all the felicitous examples in (6) ben is not modifying the propositional content of the sentence (unlike its full manner counterpart) but is rather denying a negative presupposition. In Cognola and Schifano (forthcoming) we also show that, being a discourse particle, Tr. ben does not operate at the propositional (i.e. descriptive) level of the utterance, but rather at the expressive one, where more elusive aspects of semantic-pragmatic meaning are encoded (Zimmerman 2011:2013). One piece of empirical evidence that we brought in favour of this claim was the possibility for Tr. ben to occur in imperative and yes-no interrogative clauses, without interfering with the command and question expressed, consistently with Zimmerman’s (2011:2019) remark that “[…] discourse particles are invisible to the sentence-type operators IMP[erative] and INT[errogative] […] that is, the meaning of [discourse particles] does not enter the content of the command itself, unlike all propositional material.” Note now that this test cannot be extended to ben of Group 3, which is never allowed in imperative and yes-no interrogative clauses: (11)

a.

b.

*Parla speak.IMP.2SG di andare a of go to *Sarà she.will.be quest’ora this=hour

con Maria with M. casa! (Group 3) home ben uscita di BEN left from Chiara? (Group 3) C. ben

BEN

prima before

casa a home by

Nevertheless, there are reasons to believe that the ungrammaticality of (11a), at least, follows from independent properties of ben in Group 3 and it should not be taken as an indication that ben does not operate at the expressive level in this group, i.e. its ungrammaticality correlates with the impossibility for ben in Group 3 to occur with simple tenses (cf. Restriction 2), regardless of the illocutionary force of the sentence. As for (11b) instead, which features a TAM-combination which is normally allowed (cf. epistemic future perfect), further research is required to clarify the incompatibility of discourse particles with some operators, such as INT. For the time being, we simply observe that such restriction does not only affect ben in Group 3, but also Tr. ben, which is banned with whelements and focalized constituents (see Cognola and Schifano

On the Marking of Negative Presupposition

445

forthcoming), suggesting that a wider issue may be at stake here regarding the compatibility of discourse particles with different types of illocutionary force.6 In conclusion, we will maintain the assumption that ben operates at the expressive level also in Group 3, where the same tests as those offered for Tr. ben (cf. its invisibility to INT and IMP operators) cannot be applied for independent reasons. Another fact about Tr. ben shown in Cognola and Schifano (forthcoming) is that, in this variety at least, ben is primarily connected with negation, rather than with emphatic affirmation as previously suggested for Italian. Besides the fact that ben denies a negative presupposition, this was shown by the possibility of introducing a sentence including ben with the adverb no ‘not’, while its positive counterpart ‘sì’ is ruled out (12): (12)

a.

Speaker A:

b.

Speaker B:

Battiston B. nemmeno oggi, even today

non arriverà not will.arrive secondo me according to.me (Group 1) No (/ *sì), arriverà ben NO YES he.will.arrive BEN

Interestingly, the same asymmetry seems to apply to ben in Group 3 as well (13): (13)

a.

Speaker A:

Vedrai you.will.see non porterà not will.bring mangiare eat

che Gianni that G. niente da nothing to stasera tonight

b.

Speaker B:

No (/ *sì),

avrà he.will.have qualcosa something

(Group 3) NO

YES

ben comprato BEN bought

6

Some German discourse particles, for example, are not allowed with all types of illocutionary force (cf. Doherty 1985, Zimmerman 2011: §3.2 and references cited there).

446

Chapter Fifteen

In order to give a formal account for the fact that Tr. ben is a discourse particle which is inherently presuppositional and primarily connected with negation, in Cognola and Schifano (forthcoming) we also considered its distribution with respect to sì che (lit. ‘yes that’) and no che (lit. ‘no that’), which can be used to answer a positive or negative question, with sì che appearing in a positive sentence and no che in a negative one. The examples in (14) show that Tr. ben can occur in a sì che answer to a negative question, whereas no che is excluded. The same asymmetry applies again to ben in Group 3 (15): (14)

a. b. c.

(15)

a.

b. c.

Non arriverà Battiston, quindi? (Group 1) not he.will.arrive B. then Sì che arriverà ben YES that he.will.arrive BEN *No che (non) arriverà ben NO that not he.will.arrive BEN Gianni non avrà comprato niente G. not will.have bought nothing da mangiare per stasera (Group 3) to eat for tonight Sì che avrà ben comprato qualcosa YES that he.will.have BEN bought something *No che (non) avrà ben NO that not he.will.have BEN comprato qualcosa bought something

Following Cognola and Schifano (forthcoming), we claim that sì che lexicalises PolarityP, while the adverb no in (12b)-(13b) lexicalises a higher position in the left periphery, as shown by the fact that the two can co-occur, again in both varieties: (16)

a.

b.

Battiston, B. arriverà he.will.arrive No, sì NO

(17)

a.

YES

Gianni non G. not

credo che non I.believe that not oggi (Group 1) today che arriverà that he.will.arrive avrà comprato niente will.have bought nothing

On the Marking of Negative Presupposition

b.

da to No, sì

mangiare eat che avrà NO YES that he.will.have

447

per stasera (Group 3) for tonight ben comprato qualcosa BEN bought something

The distributional facts described above are captured in Cognola and Schifano (forthcoming) for Group 1 and extended here to Group 3 by claiming that while sì che lexicalises PolarityP, no in (12)-(13) is a negative operator located in ForceP which licenses ben in the IP area. Whenever ben is present and no does not surface, we assume that high no is covertly realised in ForceP. As for ben itself, its presuppositional character and syntactic placement suggest that this should be located in NegPresuppositionalP in the lower space of the IP (also hosting mica, cf. Zanuttini 1997; Cinque 1999), from where it enters into a negative concord at the discourse level with no in ForceP: (18)

[CP [ForceP no [TopicP [FocusP [PolarityP sì che [IP [NegPresuppositionalP ben]]]]]]]

This analysis explains the superficial assertive value of ben, which follows precisely from the presence of a double negation, i.e. the one between no in ForceP and ben in NegPresuppositionalP. It also explains why ben is excluded from non-root embedded clauses (cf. Restriction 1), i.e. because it has to be licensed by an operator in ForceP, which is a typical property of discourse particles (cf. also Cinque 1999:228, fn.14 on mica; Coniglio 2008:106; Hernanz 2010:35). To sum up, we have shown that ben in Group 3 shares the same interpretative properties as Tr. ben, i.e. (i) it is an inherently presuppositional particle used to negate a negative presupposition, hence operating at the expressive rather than propositional level and (ii) it is primarily connected with negation, as shown by the fact that it can be introduced by the negative adverb no. Consequently, we can extend the same analysis to this set of varieties too, i.e. ben in Group 3 also sits in NegPresuppositionalP, where it is licensed by an operator (optionally lexicalised by no) in ForceP with which ben enters into a discourse negative concord. What remains to be done now is explaining the further restrictions which affect the distribution of ben in Group 3 but do not concern its counterpart in Trentino. The tentative claim we want to advance here is that these follow from an important intepretative difference regarding the nature of the presupposition. In Trentino ben can negate either an explicit (i.e. made explicit by a sentence in the discourse, cf. 3a in §1) or implicit (19) negative presupposition:

448

Chapter Fifteen

(19)

(addresse is taking his jacket and leaving the party) (Group 1)

(Speaker):

Ce

n’è ben ancora birra, of.it=is BEN still beer non andare via not go away “There is still some beer left, don’t leave”

LOC

The speaker’s presupposition in (19) is that the addressee is leaving because there is not beer left. This presupposition can be shared by the addresse (20a) or not (20b); in both cases the effect on information structure is that the proposition is new: (20)

a.

b.

Va bene, allora mi fermo ancora it.goes alright then myself stay another un po’ e bevo un’altra birra a little and I.drink one=other beer “Alright, I’ll stay for a while and I’ll have another beer” In realtà vado via perché sono in reality I.go away because I.am stanco, non perché non c’è tired not because not there=is più birra anymore beer “In fact, I’m leaving because I’m tired, not because there is no beer left”

On the contrary, ben in Group 3 is not compatible with out-of-theblue discourse contexts (21a), i.e. it is only felicitous if the negative presupposition is immediately recoverable in the discourse (21b): (21)

a.

(addresse is taking his jacket and leaving the party) (Group 3) (Speaker): #Dovrebbe ben esserci ancora it.should BEN be=LOC still birra, non andare via beer not go away “There should still be some beer left, don’t leave”

On the Marking of Negative Presupposition

b.

449

(A assumes that B wouldn’t have eaten the chocolate, because B does not like chocolate) (Speaker A): Li ho presi io i them I.have taken I the cioccolatini tanto chocolate.candies because a te non piacciono to you not they.like “I took the chocolate because you don’t like it” (Speaker B): In realtà, li avrei in reality them= I.would.have ben mangiati quei BEN eaten those cioccolatini chocolate.candies “In fact, I would have eaten that chocolate”

This means that, unlike in Trentino, ben in Group 3 is not compatible with a proposition adding new information to the common ground. To sum up, while the presupposition can be either explicit or implicit in Trentino, it must be necessarily immediately recoverable in the discourse context in Group 3. If we go back now to Restriction 2, i.e. mainly TAM combinations involving a non-finite form are accepted, this may be taken to follow from the fact that ben in Group 3 is incompatible with new information, i.e. a sentence with ben needs to be connected to given information and be compatible with a negative presupposition immediately recoverable from the context. These conditions disfavour the use of ben in simple tenses. Consider the following examples: (22)

a.

(Speaker A):

Li them i the tanto because

ho presi io I.have taken I cioccolatini chocolate.candies a te non piacciono to you not they.like (Group 3) “I took the chocolate because you don’t like it”

450

Chapter Fifteen

b

(Speaker B):

c

(Speaker B):

Li avrei ben them I.would.have BEN mangiati quei eaten those cioccolatini chocolate.candies “I would have eaten that chocolate” #Li mangio ben i them I.eat BEN the cioccolatini chocolate.candies “I do eat chocolate”

In Trentino (22c) would be acceptable if the speaker adds new information (23c), a strategy which is not available to Group 3: (23)

a.

(Speaker A):

b.

(Speaker B):

c.

(Speaker B):

Prendo I.take tanto because

io I a te to you

i cioccolatini the chocolate.candies non piacciono not they.like (Group 1) “I take the chocolate because you don’t like it” Li avrei ben them= I.would.have BEN mangiati quei cioccolatini eaten those chocolate.candies “I would have eaten that chocolate” Li mangio ben oggi them= I.eat BEN today perché faccio uno because I.do a strappo alla regola break to.the rule “I eat it today because I’m breaking the rules”

The examples in (22) indicate that in Group 3 ben must take scope over the whole VP, which contains the given information expressed by the presupposition, which is shared by the speaker and the hearer (cf. eating chocolate). Conversely, Tr. ben can either have scope on the whole VP

On the Marking of Negative Presupposition

451

containing the given information expressed by the shared presupposition (cf. eating chocolate) or on a single constituent expressing new information (cf. today in 23c). This asymmetry is linked to the semantic-pragmatic properties of Tr. ben which (i) can introduce new information, and (ii) is compatible with a non-anaphoric presupposition (Cognola and Schifano forthcoming). These facts are reminiscent of the properties of the German (henceforth Germ.) discourse particle doch (Egg and Zimmermann 2012), which is ambiguous between two homophonous forms and whose semantic-pragmatic properties are very similar to Italian ben (2015).7 Although semantically both forms of doch are anaphoric elements negating a negative presupposition (cf. also It. ben), they differ in their phonological and pragmatic properties, i.e. while doch1 is unstressed and does not need the proposition in which it appears to be given and backgrounded, doch2 is stressed and requires the proposition to be given and backgrounded (Egg and Zimmeramnn 2012). The similarities between the double nature of Germ. doch and It. ben are striking and may help us in understanding the differences that emerged among the different diatopic varieties. While Trentino seems to exhibit two forms of ben, corresponding to doch1 and doch2, respectively, Group 3 seems to instantiate only one form of ben, i.e. the one corresponding to stressed doch2. Pending further research on the phonological properties of ben, this parallelism is confirmed by discourse-related data. Given that Italian does not allow two focussed constituents per clause (Calabrese 1982), the fact that Trentino, but not Group 3, can also rely on an unstressed ben provides a straightforward explanation to the fact that Tr. ben can have scope on new information (i.e. a stressed constituent). This analysis also explains why compound tenses favour the semantic and pragmatic conditions for the distribution of ben in Group 3. The same reasoning applies to restructuring verbs, i.e. mood-related potere and volere are preferred over aspect-related ones like smettere (cf. Restriction 3) because of the presuppositional interpretation of ben, which is tightly connected with a modal interpretation.8 The preference for transitives over unaccusatives (cf. Restriction 4) remains instead puzzling.

7

We thank Roland Hinterhölzl (p.c.) for drawing our attention to the German facts. Another relevant factor may be the fact that smettere selects a different type of non-finite complement, i.e. an infinitive introduced by di ‘of’, as opposed to the bare infinitive selected by potere and volere (e.g. Gianni smette di fumare ‘Gianni stops smoking’ vs Gianni può / vuole mangiare ‘Gianni can / wants to eat’). Note instead that the position of the tested restructuring verbs in Cinque’s (1999) hierarchy does not seem to be relevant, as volere, which is the highest, is less

8

452

Chapter Fifteen

4. Conclusions In the present paper we have extended Cognola and Schifano’s (forthcoming) investigation on the syntactic distribution and semantic/pragmatic interpretation of Tr. ben to other regional varieties of Italian. Leaving aside the transitional area represented by Group 2 we have found that speakers located outside the productive isogloss (cf. Trentino) behave surprisingly homogeneously in their use of this discourse particle, both in terms of syntactic placement and TAM combinations admitted. Despite their geographical scatteredness (recall that Group 3 includes speakers from Friuli Venezia Giulia, Lombardy, Piedmont, Emilia Romagna, Lazio, Marche and Puglia), these speakers do not accept ben in embedded non-root contexts and with TAM combinations not involving a non-finite form; among restructuring verbs, they accept potere more readily than volere, while smettere is largely ruled out, and exhibit a preference for transitive over unaccusative verbs. As for its interpretative properties, we showed that ben is a presuppositional element employed to negate a negative presupposition, just like its Trentino counterpart. As with the latter, we accounted for this fact by showing that ben is hosted in NegPresuppositionalP from where it is licensed by a negative operator sitting in ForceP (and optionally lexicalised by no). However, we have shown that there is a fundamental interpretative difference between the two groups, i.e. ben can only occur in propositions with are given and backgrounded in Group3, whereas such a restriction is not exhibited by Trentino ben. We took this difference to correlate to the fact that ben in Group 3 can only have scope on the VP expressing the presupposition, while Tr. ben can take scope on both given and new information. By comparing such a behaviour of ben across the two groups with that of Germ. doch, we suggested that there exist two homophonous forms of ben in Italian, which correspond to unstressed doch and stressed doch. While Trentino can avail itself of both forms, Group 3 only exhibits stressed ben.

References Baardewyk-Resseguier, J. van, 1991, “Les particules de modalité wel et bien. Une approche contrastive néerlandais-français”, Cahiers de Lexicologie, 59, 39– 49. Battllori, M., M.-L. Hernanz, 2013, “Emphatic Polarity from Latin to Romance”, Lingua, 128, 9–30. accepted than potere, which is the lowest, but smettere, which lexicalises a position between the two, is largely ruled out.

On the Marking of Negative Presupposition

453

Belletti, A., 1990, Generalised Verb Movement, Turin, Rosenberg & Sellier. Belletti, A., 1994, “Verb positions: evidence from Italian”, in: D. Lightfoot, N. Hornstein (eds), Verb Movement, Cambridge, Cambridge University Press, 19–40. Calabrese, A., 1982, “Alcune ipotesi sulla struttura informazionale della frase in italiano e sul suo rapporto con la struttura fonologica”, Rivista di Grammatica Generativa, 5, 65–115. Cardinaletti, A., 2011, “German and Italian Modal Particles and Clause Structure”, Linguistic Review, 28, 493–531. Cinque, G., 1991, “Mica: Note di sintassi e pragmatica”, in: G. Cinque (ed), Teoria linguistica e sintassi italiana, Bologna, Il Mulino, 311–323. Cinque, G., 1999, Adverbs and Functional Heads, Oxford, Oxford University Press. Cinque, G., 2006, Restructuring and Functional Heads, Oxford, Oxford University Press. Coniglio, M., 2008, “Modal Particles in Italian”, University of Venice Working Papers in Linguistics, 18, 91–129. Cognola, F., N. Schifano, 2015, Italian ben: a case study on Romance discourse particles, paper presented at the 15th International Conference of the Department of Linguistics, 27-28 November 2015, University of Bucharest and Going Romance 29, 10-12 December 2015, Radbound University. Cognola, F., N. Schifano, forthcoming, “On ben in Trentino regional Italian”, Romance Languages and Linguistic Theory 2015, Amsterdam, John Benjamins. Doherty, M., 1985, Epistemische Bedeutung, Berlin, Akademia Verlag. Egg, M., Zimmermann, M., 2012, “Stressed out! Accented discourse particles – the case of DOCH”, in: A. Aguilar, A. Chernilovskaya, R. Nouwen (eds), MIT Working Papers in Linguistics, 225–238. Hernanz, M.-L., 2010, “Assertive bien in Spanish and the Left Periphery”, in: P. Benincà, N. Munaro (eds), Mapping the Left Periphery, Oxford, Oxford University Press, 19–62. Ledgeway, A., forthcoming, “The verb phrase: VP-related functional projections”, in G. Longobardi (ed), The syntax of Italian, Cambridge, Cambridge University Press. Lonzi, L., 1991, “Il sintagma avverbiale”, in: L. Renzi, G. Salvi (eds), Grande grammatica italiana di consultazione, vol. 2°, Bologna, Il Mulino, 341–412. Waltereit, R., U. Detges, 2007, “Different functions, different histories. Modal particles and discourse markers from a diachronic point of view”, Catalan Journal of Linguistics, 6, 61–80. Weydt, H., 1969, Abtönungspartikel: Die deutschen Modalwörter und ihre französischen Entsprechungen, Bad Homburg, Gehlen. Zanuttini, R., 1997, Negation and Clausal Structure, Oxford, Oxford University Press. Zimmermann, M., 2011, “Discourse particles”, in: K. von Heusinger, C. Maienborn, P. Portner (eds), Semantics (= Handbücher zur Sprach- und Kommunikationswissenschaft HSK 33.2), Berlin, Mouton de Gruyter, 2011– 2038.